1

Virtues and Vices in Scientific Inquiry

Cédric Paternotte, Milena Ivanova

(Forthcoming in Synthese – penultimate draft, January 2016)

Abstract:

The role intellectual virtues play in scientific inquiry has raised significant discussions in the

recent literature A number of authors have recently explored the link between virtue

epistemology and philosophy of science with the aim to show whether epistemic virtues can

contribute to the resolution of the problem of theory choice This paper analyses how intellectual

virtues can be beneficial for successful resolution of theory choice. We explore the role of

virtues as well as vices in scientific inquiry and their beneficial effects in the context of theory

choice. We argue that vices can play a role in widening the set of potential candidate theories

and support our claim with historical examples and normative arguments from formal social

epistemology. We argue that even though virtues appear to be neither necessary nor sufficient

for scientific success, they have a positive effect because they accelerate successful convergence

amongst scientists in theory choice situations.

1. Introduction

There has been a rising interest in the role intellectual virtues of agents play in science. A

number of authors have recently explored the link between virtue epistemology and

philosophy of science with the aim to show whether epistemic virtues can contribute to the

resolution of the problem of underdetermination (e.g. Stump (2007), [Redacted for

anonymity], Fairweather (2011), Axtell (2014)). This paper analyses how intellectual virtues

can be beneficial in the context of theory choice. Motivated by recent results from formal

social epistemology and by historical examples, we explore the role of virtues as well as vices

in scientific practice and the nature of their beneficial effects in this context. We argue that

these effects are better understood at the social level of scientific communities than at the

level of individual scientists. Vices can play a role in widening the potential candidate

theories and may favour convergence on a theory in a scientific community. However, even if

virtues appear to be neither necessary nor sufficient for scientific success, they have a positive

effect because they accelerate successful convergence amongst scientists in theory choice

situations.

The structure of this paper is as follows. Section 2 introduces some recent works in virtue

epistemology and philosophy of science that emphasise the role of intellectual and moral

virtues in science, especially in the context of theory choice. In section 3, we clarify the

notion of intellectual virtue, offer a list of virtues that are relevant for theory choice and

identify one of their common points. Section 4 presents cases of beneficial vices in scientific

inquiry. Section 5 offers normative reasons for the usefulness of intellectual vices in science

by introducing recent models in social epistemology that emphasise the positive roles of self-

interest, conformism and dogmatism for scientific success. Together, section 4 and 5 make the

case that vices play an important role in scientific progress. In sections 6 and 7 we establish

the positive role for virtues. Section 6 introduces two kinds of scientific activity and

highlights the different virtues and vices beneficial in each context. Section 7 identifies a

positive role that epistemic virtues can play in theory choice: while neither necessary nor

sufficient, we argue that epistemic virtues accelerate convergence in a scientific community.

Section 8 defends the proposed account of virtues against possible objections.

2

2. Virtues in epistemology and philosophy of science

Intellectual and moral virtues of agents have recently become prominent in the analysis of

knowledge with the development of virtue epistemology, according to which knowledge

should be understood in terms of the intellectual and moral virtues of agents who achieve

epistemic success.1 Several authors have recently applied virtue epistemology to issues in

philosophy of science in order to highlight the role virtues play in scientific inquiry. For

example, David Stump (2007) revives Pierre Duhem’s concept of ‘good sense’ in order to

show the relevance of virtues to the problem of theory choice.2 Duhem (1906) famously

argues that in situations where neither experiment nor logic guides scientists in their decisions,

they employ moral and intellectual virtues to make a choice between alternative theories.

Stump highlights important similarities between virtue epistemology and Duhem’s appeal to

‘good sense’ in theory choice. First, virtue epistemologists appeal to similar virtues –

impartiality, sobriety, honesty – in the context of knowledge acquisition to those used by

Duhem in the context of theory choice. Second, both Duhem’s ‘good sense’ and the virtue

epistemological account of knowledge are non-algorithmic. It has been recognised that

intellectual and moral virtues play significant role in Duhem’s resolution of theory choice and

in their recent, [redacted for anonymity] show how good sense functions at the level of the

scientific community in order to favour consensus formation in situations of theory choice.3

The important idea behind these recent works is that much scientific activity – whether

it is resolution of theory choice or consensus as to what experiment or observation to make –

can be characterised as involving the exercise of epistemic virtue.4 How are epistemic virtues

understood within the virtue epistemology debate? Virtue epistemology can be divided into

two main groups: reliabilist and responsibilist. Reliabilists analyse knowers in terms of

cognitive abilities such as reliable memory, perception, good reasoning, etc. (Greco 1999),

which they value because they are supposed to be truth-conducive. Responsibilists, on the

other hand, focus on character traits such as intellectual courage, open-mindedness,

intellectual honesty, sobriety, impartiality (Baehr 2011). These intellectual virtues are not

taken to be reliable or truth-conductive because in unfavourable environmental conditions the

virtues cannot guarantee the production of true beliefs (Zagzebski 2003). Now, the virtues

mentioned in the context of theory choice are typically the ones emphasised by responsibilist

virtue epistemologists. 5 Accordingly, in this paper we will focus on intellectual virtues

discussed by responsibilist virtue epistemologists and explore their role in situations of theory

choice, while staying neutral between the reliabilist and responsibilist approaches. We will

adopt a weak understanding of epistemic success; rather than understanding success in terms

of true theories, we will take it to be temporary empirical adequacy. Our aim is to identify the

effects virtues can have on a specific scientific activity – the resolution of theory choice. As a

consequence, we will defend an account that highlights the positive effect of virtues at the

1 In such accounts, knowledge is defined as a state that includes both the true belief and the virtuous source

responsible for the acquisition of this belief. This is taken to block the Gettier-cases (1963) without adding an

extra condition on knowledge. As Greco (2002) notes, the Gettier cases are blocked once we draw the distinction

between (1) a belief being true and being the result of epistemic virtues; and (2) a belief being true because of it

being the result of epistemic virtues. 2 See Fairweather (2011) and [Redacted for anonymity] for a further discussion of Stump’s interpretation. 3 The role of virtues has been further explored by Axtell (2014) and [Redacted for anonymity] in the context of

theory choice; by Morton (2014a), Vallor (2014) and Di Bucchianico (2014) in the performance of experiments

and observations. 4 Not to mention that scientific activity itself has occasionally been seen as favouring the cultivation and exercise

of virtues, as pointed out by an anonymous referee. 5 A point already made by Kidd (2011), who interprets Duhem as a virtue responsibilist.

3

social context.6 We first consider what intellectual virtues are relevant in the context of theory

choice. We then analyse the role of epistemic vice as exemplified by certain historical

examples and motivated by recent models in social epistemology. This will allow us to

delineate the beneficial effects of epistemic virtues in situations of theory choice.

3. Intellectual virtues in theory choice

What virtues are relevant to a scientific context? Our focus is on intellectual and moral virtues

and their epistemic effects. We will use the term ‘intellectual’ virtues to cover both meanings

as we do not take the distinction between intellectual and moral virtues to be a sharp one.7,8

It is crucial to note that some virtues are relevant to theory choice without being

manifested during decision-making. Consider creativity or imagination: they are intellectual

virtues from which individuals come up with new ideas (which may or may not become

competitors later). In the context of theory choice, let us call them ‘generation virtues’, in

opposition to ‘assessment virtues’, that influence the choice of a theory among existing

candidates. This distinction illustrates a way in which different virtues could be relevant to

different kinds of scientific activities. In other words, a given virtue may not be beneficial or

detrimental to science in general, but rather to specific kinds of scientific activity, such as

theory creation or theory choice.9 In what follows, we focus on virtues relevant to theory

choice situation and exclude from our discussion ‘generation virtues’, such as imagination and

intuition, which are relevant in the conception of a new theory. The reason for this exclusion

is that in the generation phase agents are not committed to the truth of the theory and have not

necessarily formed beliefs about it.

Let us now provide brief characterisations of the main virtues found relevant to theory

choice situations in the literature, namely: open-mindedness, intellectual courage, intellectual

sobriety, intellectual humility, faithfulness, integrity, disinterestedness, honesty and

impartiality. This list is deliberately long, lest we seem to have chosen ad hoc virtues that fit

our purpose. However, we intend it to be neither exhaustive nor set in stone, because the set

of relevant intellectual virtues is likely to be influenced and constrained by the historical,

social and institutional contexts (Daston & Gallison 2007, Shapin 2008).

The following characterisations are neither supposed to cover every intuitive meaning

of each virtue nor to fully define them, but aim to be minimal so as to avoid as much overlap

as possible,10 to emphasise the distinctiveness of each virtue and focus on their effects on the

dispositions of individuals.

6 Recent works on the social dimension of virtues and vices are Rowbottom 2011, [Redacted for anonymity] and

Morton (2014b); and Kidd (2014) in the case of intellectual courage (also see Roberts and Wood 2007: Ch.5). In

virtue epistemology, a similar claim has been made by Kawall (2002). Fricker (2012) has provided an influential

social virtue epistemology approach to group testimony. However, even though she holds that groups can

possess the virtues of good informants, she does not discuss how the virtues of individual members impact

collective epistemic success. 7 Zagzebski (2003) and Putnam (1981) consider moral virtues to be fundamental. They appeal to the Aristotelian

notion of eudemonia, the idea of personal flourishing. 8 Intuitively, moral and intellectual scientific virtues may seem quite distinct: a good researcher could be

immoral, and a bad researcher morally righteous. However, this intuition equates the quality of research with the

intellectual virtues it arises from. By contrast, we hold that some virtues considered as moral can influence the

quality of a scientist’s output. 9

An analogous distinction is introduced in Sosa (2011) who distinguishes between ‘transmission’ and

‘generation’ faculties. 10 We do not mean to presuppose that virtues should not overlap, as some authors have argued for their similarity

or even unity (Baehr 2011: ch.1). However, we intend to focus on their differences, so that the common point we

later highlight does not appear built-in.

4

Open-mindedness (Zagzebski 2003) guarantees that agents do not exclude some

candidate hypothesis or theory on a priori grounds and are able to consider the merits of

different alternatives at all times regardless of their background beliefs and values.

Hypotheses are still refutable: open-minded agents only refuse to reject hypotheses before

considering the evidence.

Honesty, extensively discussed in Kawall (2002), is also called rectitude and probity

by Duhem (1906, 218). Honest scientists do not withhold, destruct, alter or forge results: they

make the totality of their evidence or results available to the scientific community. Honest

scientists also do not defend or attack hypotheses or theories with arguments they know

fallacious.

Faithfulness or loyalty, again discussed by Duhem and Zagzebski, implies that agents

have the tendency to cling to the theory, methods or background beliefs of their research

group (or their own) because they or their group favoured them in the past. Although loyalty

is usually seen as a virtue, it is not necessarily so in a scientific context. In particular it seems

to contradict intellectual courage, which may require a separation from one’s group if

necessary, or to intellectual humility in the individual case. Loyalty may have positive effects

when the scientist (or her group) happens to work on a theory that will turn out to be

successful.

Intellectual courage (Montmarquet (1993)) makes agents maintain their beliefs,

interests, methods, values, no matter the number of agents that hold different ones. Of course

courageous agents can still be influenced by others’ beliefs or results when evidentially

relevant. Intellectual courage can also mean that agents’ decisions are not influenced by

expected losses in terms of credits, reputation, etc. For Zagzebski (2003), it includes

perseverance and determination as well.

Intellectual humility is a virtue that has received a lot of attention in the recent

literature (Robert and Wood 2007). It is typically opposed to arrogance, vanity or hubris. We

understand it as an absence of self-centred bias. Humble scientists are likely to ignore their

personal interests or values and not to exaggerate the relevance or significance of their results.

They are also more likely to recognise their limitations and acknowledge problems for the

theory they develop.

Intellectual integrity, investigated in detail in Zagzebski (1996) and Kawall (2002), is

the virtue that tends to make scientists make decisions independently of their expected

rewards (other than scientific success, understood as the desire to obtain empirically adequate

theories). Integrity implies that personal gain and conflicts of interest cannot influence the

scientist’s choices. As is immediately clear, a claim that this virtue is crucial for scientific

success would clash with some prevalent analyses of the social structure and reward system of

science. According to Robert Merton (1957), for example, priority and credit are the driving

factors behind the scientific enterprise, which plays a positive role because it secures

optimality.11 It is also at odds with Kitcher’s (1990) and Strevens’ (2003) analyses, which

show that personal interest is perfectly compatible with efficient science.12

Impartiality is another virtue that virtue epistemologists such as Montmarquet and

Zagzebski 13 regard as essential for truth acquisition. Understood in a narrow sense,

impartiality means that prior beliefs about theories are equally strong. More broadly construed,

impartiality implies that scientists’ decisions are not influenced by the source of theories or

the source of the evidence (such as the community, ethnicity, religion, gender of its author(s)).

In particular, impartiality entails intellectual humility.

11 Hull (1988) also famously highlighted the crucial role of credit in science. 12 Their views are discussed with more details in section 5. 13 But also Duhem (1906), who saw it as independence from metaphysical commitments.

5

Disinterestedness echoes Duhem’s ‘detachment from all interests and passions’. It can

be understood as the lack of influence of any interests, 14 or self-centred bias, with the

exception of those underlying scientific inquiry and its good practice – which motivate theory

choice in the first place.15 It is entailed by the conjunction of impartiality and integrity.

Finally, intellectual sobriety, described as ‘anti-enthusiasm’ (Zagzebski 2003) or as

detachment (Duhem 1991, also cited in Stump 2007), is difficult to define but can also be

understood broadly and narrowly. In the narrow sense, sobriety implies the lack of influence

from the expectations of the theory’s consequences. In a broader sense sobriety implies that

the agents are not influenced in their decision by anything other than evidence.

Except for faithfulness (which we exclude for already stated reasons) and honesty, the

assessment virtues share the following crucial characteristic: they suppress reasons for theory

choice that are based on factors other than evidence, including reasons that restrict evidence

or distort its confirmatory power.16 Even open-mindedness is a particular case that prevents

such reasons from downright eliminating candidate hypotheses in the absence of evidence. In

short, most virtues suppress non-evidential reasons in theory choice. (In that sense they may

be gathered under the broad definition of intellectual sobriety.) One may be tempted to

conclude that this observation makes the positive role of virtues in science clear – for what

could make a scientist more successful than the dispassionate and exclusive care for

evidence? However, the next sections will make us realise that science has benefited and even

should benefit from certain vices. There will be a positive role left for virtues in science, but a

subtler and more modest one.17

4. Historical examples of beneficial vice

Linking scientific success to the virtuous character of scientists appears at first glance

descriptively incorrect. The history of science is replete with examples suggesting that

scientific success may often stem from vices of different kinds rather than virtues. In this

section we describe two cases of the beneficial effects of intellectual vices, one at the

individual and one at the community level.

Louis Pasteur, one of the most significant scientists of the 19th century, performed a

series of experiments during the 1860s, in which he tried to show that life cannot

spontaneously be generated without the presence of a parent organism (Waller 2002). Felix

Pouchet and Charlton Bastian had performed numerous experiments that appeared to confirm

the theory of spontaneous generation – experiments in which a sterile infusion not

contaminated by atmospheric air gave rise to the life of new microorganism. Pasteur publicly

claimed that his opponents’ experiments were flawed – that the sterilizing process was not

14 Note that this does not exclude interests but their influence. Stump’s interpretation of Duhem’s views is that if

scientists can have commitments or “ethical and political values”, they “cannot be connected directly to the

scientific choice one is making.” (2007:158). 15 Here we are indebted to an anonymous referee. 16 This is not to say that virtues do not share other common points. For instance, following Aristotle, it could be

argued that having the 'right motivation' is a constitutive component of any virtue. For instance, a scientist would

not be considered open-minded if she considered the merits of different alternatives at all times out of (say)

unreflective conformity to a learned set of rules. We thank an anonymous referee for making this point. 17 As a final preliminary note, we emphasise that we do not aim to define virtues and vices in instrumental terms

– where character traits would qualify as virtues or vices depending on whether they lead to scientific success or

failure. Quite the contrary: we have extracted from the literature what are usually considered as virtues and vices

for non-instrumental reasons, in order to determine their respective effects on scientific success. In particular, in

sections 4-6, we will point out that several vices can lead to scientific success, while continuing to call them

intellectual vices. One who favours an instrumental view of epistemic character traits may want to use such

discussion to argue that the foregoing actually makes these vices virtues; but this paper does not take such a

stand.

6

done properly and thus the solution in each case was contaminated. In particular, he was

credited with showing that Pouchet had used contaminated mercury after performing a public

experiment at the Sorbonne in 1864.18 However, Pasteur’s notebooks reveal that as soon as

1861 he had gathered but kept secret evidence in favour of spontaneous generation. In

addition, in 1864 he refused to publicly replicate one of Pouchet’s mercury-free experiments

that at the time would have been taken as showing the possibility of spontaneous generation.

Years later, in 1877, he managed to prevent Bastian from performing a public experiment

aimed at supporting spontaneous generation, while accusing him of being a sloppy

experimenter. However, his notebooks reveal that he and his team secretly duplicated his

findings in order to understand how to undermine spontaneous generation.

Additional considerations suggest that Pasteur thought that the solutions had been

contaminated by some source and thus was convinced by the truth of his germ theory, despite

the fact that the available evidence was at best indecisive between the two theories. In his

notebooks, he consistently labelled experimental results as successful or unsuccessful,

depending on whether they respectively favoured germ theory or spontaneous generation.

Moreover, Claude Bernard’s notes, revealed after his death in 1878, contained claims that

Pasteur’s support of germ theory was based on bias rather than evidence. Finally, Pasteur’s

own notes later revealed that he held religious beliefs incompatible with spontaneous

generation (as new life could only be created by God) (Waller 2002, following Geison 2005).

This evidence does not prove that Pasteur was fully intellectually vicious, in the sense

of being deceptive (as could have shown a confession in his notebooks, for instance). We do

not know of any such unequivocal example of admitted vice causally linked to substantial

scientific success. Still, the previous considerations strongly suggests that Pasteur was partly

led by self-interest and bias, dishonestly withheld from the public experiments that did not

support his theory and at times refused to enter in a debate with his opponents. Instead he

used his social and political power to eventually silence them. Yet, his conviction in the germ

theory was eventually to become one of the most significant discoveries in the history of

science.19,20

Vices may not be beneficial only at the individual level; theories can be adopted

because biases and dogmatism are widespread in a community. In the beginning of the

twentieth century, Mendelism came to dominate theories of non-nuclear inheritance, which

were soon abandoned, although for non-empirical reasons. Solomon (2001) notes that among

other factors, Mendelism was favoured by a number of biases and interests: it was supported

by the eugenic movement, of which most Mendelians were part; it was similar to the then

widely accepted hierarchical views of family and society; and it was financially supported by

plant and animal breeders. Importantly, it follows from Solomon’s analysis of the various

factors involved that had these three vices been rare or absent, Mendelism would not have

dominated theories of non-nuclear inheritance (Solomon 2001, 85). In short, biases and

interests were crucial in bringing about what later turned out to be the better supported theory.

18 We now know that Pasteur’s evidence was inconclusive, because its strength relied on the premise that

microbes cannot survive exposure to boiling water. 19 Note that our aim is not to dismiss Pasteur, who was a remarkable scientist and an extremely skillful

experimenter; on the contrary, that Pasteur was scientifically skilled and successful makes his intellectual vices

and their effects all the more salient. 20 Millikan’s oil drop experiments, performed in order to measure the charge of the electron, offer another

example of beneficial dishonesty, although a more debatable one. Millikan’s deletion of one set of experimental

results had the effect of reducing the statistical error of the measurements (Franklin 1998, 429), which made the

results more persuasive and accepted with less difficulty than would have been the case from the complete set of

results. However, as in other similar cases, Millikan’s behaviour may be interpreted as exemplifying Duhem’s

scientific good sense, which may be characterised as guiding both virtuous and vicious behaviours ([Redacted

for anonymity]).

7

These examples suggest that behaviours motivated by epistemic vice have led to

epistemic success in the history of science. However, one can explain this apparent epistemic

success in terms of lucky behaviour due to favourable environmental conditions rather than

individual excellence. More often, or in the long run, we would expect dogmatic, prejudiced,

interested and dishonest behaviours to lead to bad science and lack of epistemic success. One

could also offer a contextual justification of virtues, an approach we do not adopt, and argue

that the apparent vicious behaviours were actually virtuous. The positive role of vice in

scientific practice, however, seems to be highlighted by recent results in social epistemology.

While this literature is not usually related to the debate on epistemic virtues, in the next

section we show that it offers normative reasons for the beneficial role of epistemic vice.

5. Epistemic virtue and vice in formal social epistemology

Normative reasons to consider vices useful to science can be drawn from the literature

concerned with the benefits of diversity in science. These works typically aim to investigate

when, how and why does the diversity within a scientific community contribute to its

epistemic success, through modelling and multi-agent simulations approaches. They can also

be interpreted as showing that science should benefit from the presence of various individual

vices in scientific communities.21 We focus on three such approaches, corresponding to three

different vices: egoism, conformism and dogmatism.

Kitcher (1990) and Strevens (2003) show in simple models that it may be optimal for

scientists to distribute themselves among different projects that aim to solve the same problem,

even when one has a higher probability of success; that is, science can benefit from the

division of cognitive labour among scientists. Such division of labour may emerge if scientists

are solely motivated by the pursuit of truth. However, as Kitcher, later refined by Strevens,

shows, it may also emerge when scientists choose projects depending on the expected

personal rewards (in terms of credit) they entail. As a consequence, a scientific community

composed of purely egoistic or self-interested agents would perform at least as well as one

composed of ideal truth-seekers. 22 Kitcher’s and Strevens’ results can be interpreted as

justifying the scientific benefits of egoism or interestedness. This is problematic for a defence

of the positive role of virtues in science: first because egoism and self-interestedness are

typically considered as vices; second and more importantly, because the positive role of

virtues in epistemology partly stems from the motivation to pursue truth they provide. As

Baehr (2013) notes:

What can be said about the sorts of demands or challenges imposed by successful inquiry?

One such demand is fundamentally motivational, for inquiry must be initiated or undertaken

by […] intellectual virtues like inquisitiveness, reflectiveness, contemplativeness, curiosity,

and wonder can be essential to a successful pursuit of the truth. (19)

In other words, egoism is opposed to virtues both morally and intellectually.

Second, Weisberg and Muldoon (2009) investigate diversity questions through multi-

agent simulations. They model scientists as agents exploring ‘epistemic landscapes’, in which

21 We emphasise that the interpretation of the following works in terms of vices is our own and does not commit

their authors. 22 For Kitcher and Strevens, the widespread following of epistemic norms may lead to suboptimal results in

science; pursuing personal interests can lead to collectively better outcomes. Still, in principle agents who only

care about overall scientific success would equally agree to distribute themselves among projects. In this sense,

following private interests is not strictly beneficial to science. However, it can be beneficial in practice, as the

hypothesis that scientists only care about scientific progress is descriptively inadequate.

8

hills represent peaks of ‘epistemic significance’ and investigate the kind of strategies that

make a population more likely to reach such peaks, that is, more epistemically successful.

Interestingly, the most successful populations are polymorphic, consisting in a minority of

‘maverick’ agents who favour previously unexplored parts of the landscape, and of a majority

of ‘followers’ attracted by existing successful approaches. Briefly put, mavericks explore and

followers exploit: the former allow the latter to home in on epistemic peaks.

Here again, results intended to show the benefits of diversity also reveal possible

benefits of a particular vice, namely conformism – the corresponding virtues being, for

instance, intellectual courage, creativity and imagination. Unlike the case of egoism, universal

conformism would harm scientific communities; still, to be collectively efficient communities

should contain a fair share of conformists, which makes it necessary to scientific success.

Finally, Zollman (2010) endorses the view that division of cognitive labour is

beneficial to science and identifies two ways in which it can be maintained long enough. One

way is to limit scientists’ access to information, for instance by embedding them in

communication structures such that they only know what some of their colleagues do. As a

result, different scientists are likely to have access to different information and thus to make

different research choices. Another way is to have some members of scientific communities

entertain extreme beliefs, as is the case when scientists defend theories that most find

unreasonable or misguided. Science may benefit from both diversity of available information

and epistemic diversity, that is, diversity of beliefs.

The two aspects of diversity can be realised by individual vices. The presence of

dogmatic scientists within a community could favour epistemic diversity as their beliefs

would be either extreme or resistant to problematic evidence.23 It could also favour diversity

of available information as dogmatism towards one’s pet theory could lead to ignorance or

disregard towards the activity of other teams.24 In other words, Zollman shows the benefits of

dogmatism, whether manifested in the shape of unjustified extreme prior beliefs, or of

informational limits, which can be realised by intentional blindness to specific sources of

evidence.25

Because these arguments are based on idealised models and simulations, the foregoing

does not prove that science cannot be successful in the absence of individual vices. Nowhere

in these models are vices represented in a strong sense (involving being motivated by certain

reasons and principles); all we can say is that the agents’ apparent interests and observed

behaviours could result from intellectual vices.

Still, the models can be interpreted as revealing possible consequences of intellectual

vices on scientific success, which is sufficient for our purpose. In particular, the three

examples show that it is possible for science to benefit from vices and for some epistemic

virtues not to be necessary for scientific success. Furthermore, some scientific successes have

23 This could also be realised by having virtuous members playing the role of the Devil’s Advocate in debates, in

the interest of their interlocutor or of the whole community. We take this positive role to be also social in nature,

and covered by Rowbottom’s (2011) analysis. Still, it is consistent with dogmatism having similar effects, and so

does not mitigate our later claim that virtues are not necessary for scientific success. 24 Informational diversity could also be favoured by dishonesty, which would lead scientists to hide part of their

data or results and would thus favour the occurrence of different states of information within a community. 25 Note that the benefits of dogmatism must be taken with a grain of salt, as Zollman emphasises that the

concomitant presence of both corresponding diversities is likely to be detrimental, as it would ‘hinder the

convergence to one action’ (Zollman 2010, p. 33). In Zollman’s words, diversity only helps if it is ‘transient’,

that is, temporary. If not, scientists may never become able to converge on the best theory. Under the ‘vice’

interpretation, this means that extreme dogmatism (or its widespread combination with dishonesty) would be

harmful to science. Non-formal analyses reflect this equivocal effect: for Stanford’s (forthcoming), the social

structure of contemporary science has led to a detrimental degree of conservatism; however, Rowbottom (2011)

argues that dogmatism can be a functional feature of science at the group level.

9

stemmed from vices, and some kinds of scientific successes can be justified by the presence

of such vices. In other words, we have cast descriptive and normative doubt on the intuitive

idea that scientific communities cannot be successful if not composed only of intellectually

virtuous agents.

Before we turn to the positive role of virtues in science, note that a change of

perspective has occurred during the transition between descriptive and normative arguments.

In our historical cases, vices improved the epistemic success of individuals in isolation:

because of their biases, dogmatism or dishonesty, some individuals were able to formulate

and defend neglected, discredited or new hypotheses or theories.26 By contrast, in this section,

vices are only beneficial because of their structural effect on the epistemic success of a

community: conformist or dogmatic individuals would not necessarily have been better off in

isolation.27 In other words, we have now entered the field of social epistemology. Indeed, the

positive role of virtues that we will ultimately identify necessitates the presence of a social

context.

6. Pre- and post-convergence scientific activity

Given that vices can provide scientific success, an elaboration of the beneficial role of virtues

needs to be given. How could science benefit from virtues? Unfortunately, there is no general

answer to this question. This is because science involves different kinds of activities relevant

to theory choice, for which specific virtues or vices play different roles.

In principle, in the context of theory choice, scientific activity can be seen as a three-

step process. First, in the absence of any satisfactory background theory, numerous theories

are created, built, developed and tested by individual members of the scientific community (or

small teams). Then, some of these communities reach a consensus on a candidate theory; as

research teams, they start further investigating and refining these theories. Finally, at some

point, the theory becomes less successful and loses momentum; the community dissipates and

each member may start working again on new candidates. Briefly put, theories are created, are

explored or developed, before they decline. Each of these steps can be brief or long. Some

theories originate from a unique, sudden illumination; others from a long process of trials and

errors. Theory development can be short or long depending on how fast critical difficulties are

encountered. A theory’s decline can be fast or slow depending on the number and quality of

alternative theories.

These three conceptual steps actually correspond to only two kinds of scientific

activities. Whether a theory is fruitfully developed or declines does not depend on what

scientists do but on the results they obtain. In both cases scientists keep trying to test and

modify the theory, either to expand its scope or to deal with anomalies. Theory creation is

distinct though, as is frequently revealed by the lack of scientific methods for building or

coming up with a promising theory in the first place.

As a result, theory development comprises pre-convergence and post-convergence

phases. In the pre-convergence phase, theories are created and elaborated to some extent.

There is, however, no agreement in the scientific community as to what theory is most

promising or worthy of pursuit. This can result either from the absence of confirming

evidence, or by the lack of agreement as to what counts as evidence for or against theories. In

the post-convergence phase, a certain amount of consensus as to what theories are most

26 Strictly speaking, as noted in the previous section, the case of Mendelism has a social dimension. However, it

involves aggregated individual vices, the effect of which stems from their sheer number. By contrast, in the

normative cases, the benefits of vices depend on the social context. 27 In Kitcher’s and Strevens’ models though, egoistic individuals could have been better off in isolation.

10

promising has emerged. Some theories are refined, more systematically tested, possibly

expanded and possibly abandoned at some point. In other words, the post-convergence phase

involves collective selection and exploration of theories.

Overall, the product of science can be described as a constantly changing set of

theories, on which scientists are likely to aggregate for a while. As long as a theory is

moderately successful, aggregation around it increases or is at least stable; when too many

difficulties arise, it decreases and the theory may even disappear when no one works on it or

believes in it anymore.

This picture of science is general enough to fit the reader’s favourite view of scientific

change. Pre-convergence may correspond to revolutionary science, to the emergence of

research programs, to a context of discovery. Post-convergence science may correspond to

normal science as puzzle-solving activity, to the development and decline of research

programs, to the pursuit of theories, to a context of justification. The picture is agnostic about

the number of available theories and the degree of dominance that any theory enjoys at a

given moment.

Indeed, we do not intend the talk of pre- and post-convergence to be groundbreaking.

To say that theories appear, stay for a while and disappear is almost vacuous. However, this

has important consequences for the role of virtues. This is because different virtues play

different roles in pre- and post-convergence. Consider post-convergence science first. Are any

virtues helpful when a group is to explore and refine a given theory or research program? As

is well known since Kuhn, scientists who work in a given theoretical framework can be

partial; biases towards one’s framework and against alternative ones are possible and even

useful, as they allow deeper refinements of the theory. Post-convergence scientists can also be

interested in all sorts of ways, as Kitcher and Strevens teach us. A certain form of honesty

should be observed – there is no harm into communicating possibly damaging results if

everyone is identically inclined in favour of the theory. Loyalty among members should be

positive as well, as it ensures the stability of such similar inclinations. In any case, post-

convergence science is certainly possible for interested, partial, narrow-minded, conformist

individuals – as long as these vices are similarly oriented. As a consequence, virtues are not

necessary for post-convergence science.

Virtues are even less useful in pre-convergence science, where all that matters is that a

high enough number of different theories are created. Be intellectually arrogant, biased or

interested - whatever helps you create your own unique pet theory is welcome. This is a phase

for Fereyabendian proliferation, where ‘anything goes’.28 Even honesty is not necessary: to

communicate only the most favourable results may help the theory survive longer;

shortcomings will be scrutinised soon enough once the interest of others is piqued. Recall our

examples in section 4: thanks to Millikan’s dishonesty, Bohm’s dogmatism and Pasteur’s

biases, the scientific community came to entertain hypotheses and theories that turned out to

be successful, or at least reasonable alternatives to the established ones. Most vices (e.g.

dogmatism, dishonesty, partiality, arrogance, interestedness) favour theory diversity, which in

turn impacts the probability that a (later to be) successful theory has been discovered.

There is still room for virtues in pre-convergence science though, although of a

different kind. Recall the distinction between generation and assessment virtues. Pre-

convergence science may be compatible with a few assessment virtues, such as intellectual

courage (the ‘good’ dogmatism) and open-mindedness (we do not want scientists to eliminate

candidates at this stage). By contrast, generation virtues are immensely beneficial to pre-

28 Which Feyerabend intended only as a mock methodological principle (Tsou 2003), meaning the absence of

any such principle; this fits theory creation during pre-convergence science.

11

convergence science, as creativity and imagination naturally favour the apparition of new

candidate theories.29

This is a bleak picture for assessment virtues: not only do they differ in pre- and post-

convergence science, but they seem unnecessary. However, this conclusion is premature as it

neglects what happens between the two phases. Indeed, our claim is precisely that virtues

favour the transition between pre- and post-convergence phases.

7. The positive role of intellectual virtues in science

7.a. Virtues and convergence

The distinction between pre-convergence and post-convergence activities is clear. However, a

theory’s life may not exhibit both. No theory is guaranteed to reach the post-convergence

phase: it may fail to attract enough attention, whether for good or bad reasons. We claim that

this is where virtues come into play with an unambiguous positive role. Crucially, this role

only occurs in social contexts, when a sufficient number of scientists exemplify assessment

virtues.

What exactly happens during convergence on a theory? During pre-convergence,

scientists disagree as to the merits of various theories; during post-convergence, some

scientists have agreed and gathered to jointly explore it. So convergence happens partly

because a number of scientists come to agree on a theory’s merits or share a similar judgment.

Convergence is more likely to happen when similarly inclined minds meet, or when

dissimilarly inclined minds reach a consensus. In terms employed in the diversity literature,

convergence may happen when scientists’ judgments of epistemic utility or epistemic

significance are aligned. Most diversity remains as division of labour; scientists share similar

standards and part of their goals.30

Even during transitions, the role of virtues is not obvious. They are not necessary for

convergence. Identically biased, partial scientists would naturally converge on the same

theory – groups of similarly prejudiced people are commonplace even in non-scientific

contexts – a lesson also drawn by Solomon (2001) and Miller (2013). They emphasise that

consensus can emerge accidentally or for bad reasons, as repeatedly observed in the history of

science (Solomon 2001, 121-135). As seen with Mendelism, bad reasons may even have

favoured what retrospectively appears as the most justified available theory.

Maybe virtues are at least sufficient for convergence? Surely a group of impartial,

disinterested, sober, courageous, honest scientists would choose to explore the same theory;

untainted by biases, private interests and properties of its sources, they would consider the

same evidence, assess is similarly and then, as evidence accumulates, focus on the same

theory (if on any). We have seen that most virtues tend to suppress the influence of ‘bad’

reasons for theory choice. They render the assessments of confirmatory relations between

evidence and theories of all virtuous scientists more similar, and their choices based on

evidence alone. Even if they obtain diverse results, use different methods and favour different

29 The positive role of generation virtues for pre-convergence science looks obvious, as the pre/post convergence

distinction seems to echo the generation/assessment virtue one. However, the parallel is not artificial. First,

virtue epistemologists and philosophers of science developed these independently. Second, post-convergence

science is not straightforwardly made more successful by assessment virtues. 30 We do not claim such convergence to be universal but only partial. We agree with Solomon’s (2001) claim

that dissent may be appropriate if properly distributed and that theories’ power of attraction should be

proportional to their degree of empirical and technological success. In such cases, scientists still agree about the

theories’ merits, even if other factors influence their choice to work on a specific one. Partial convergence still

stems from agreement.

12

theoretical virtues, scientists who share virtues are more likely to converge on a theory. In

other words, scientific virtues increase the chances that convergence takes place – that

research communities solidify around a theory and adopt it as a paradigm, research program,

etc.

Does this distinguish virtues from vices? It is hard to say, as the effects of vices on

convergence are difficult to assess. Solomon analyses several historical cases of consensus in

which non-empirical factors were instrumental; to which Tucker (2003) replies that consensus

appears unlikely in the presence of a variety of individual biases, as, similar vices are less

likely to have similar effects than similar virtues do. For instance, a community of arrogant

scientists, each favouring their own pet theory, would be unlikely to manifest convergence.

By contrast, as most virtues entail the absence of specific reasons to choose theories, they

reduce the opportunities for choice diversity.

Miller objects that ‘it is plausible that different biases would all pull at the same

direction’ (2013:1307). However, Miller supports his view by one historical example (peptic

ulcers) and the claim that ‘when people and groups have a variety of interests and hold views

on many issues, they are almost bound to agree on occasion on some things’ (Ibid.,1307-08),

from which conclusions regarding likelihood cannot be drawn. On balance, the effect of a

variety of vices on consensus seems to be on the negative side. However, what matters is not

mere convergence but convergence on successful theories, to which we now turn.

7.b. Virtues and successful convergence

Our starting question is whether virtues contribute to scientific success, defined in terms of

adoptions of temporarily empirically adequate theories. Mere convergence does not guarantee

success; the theory on which scientists converge must also be empirically adequate – lead to

new predictions or explanations of existing phenomena. A community of ideally virtuous

scientists may converge too fast, before anyone has time to come up with a decent theory.

This echoes the question - deemed crucial by Duhem – of the right timing for theory choice:

choice must be based on the right parameters, but also occur only when a good theory is

available.31

We have seen that similar vices also facilitate convergence; why should they make

successful convergence less likely than virtues? At first glance, the odds that non-virtuous

scientists converge on a good theory may seem low. Scientists equally prejudiced in favour of

an actually promising theory will be successful; however, if their bias stem from vices –

uncorrelated with possible success – they would have been unsuccessful in most possible

worlds (in which the most successful theory would be different). By contrast, virtues, which

lead to evidence-based convergence, should result in higher average success across possible

worlds.32 In particular cases, virtues may be less efficient than dogmatism – with hindsight, it

always appears faster not to err on the side of theories that end up failing – but they are more

efficient on average. Unsurprisingly, favouring evidence leads to the most successful theories.

There are other reasons why a virtue-based consensus is likely to form around good

theories. Miller (2013) identifies three possible causes for bad (not knowledge-based)

consensus: ‘veritic luck’ (Gettier cases), ‘epistemic misfortune’ (systematic misleading

factors), and ‘non-cognitive reasons’ (e.g. political, material, reputational…) (ibid., 1299). He

then provides two conditions for consensus to be truth-conducive: agents have to share the

same background assumptions, such as their formalism, ontological schemes and evidential

31 [Redacted for anonymity]. 32 This would be true even if all instances of past successes crucially depended on vices. Actual success may be

coincidental. However, the probability that this keeps being coincidental should decrease as time passes, due to

the law of large numbers.

13

standards (‘social calibration’); and must form their views independently, both in terms of

evidence-gathering techniques (‘apparent consilience of information’) and of perspectives

(‘social diversity’).

The latter requirement that agents form their views independently plays a similar role

as generation virtues do: it makes the presence of good (true, or temporarily empirically

adequate) theories in the choice set more likely. Note that background diversity, although

crucial, is not sufficient if hampered by uniformising mechanisms (ethnic diversity is useless

when racist scientists strongly influence theory choice). Assessment virtues precisely prevent

such uniformising mechanisms: they eliminate non-evidenced based reasons for theory choice

based such as characteristics of the theory production.

As for the former requirement of social calibration, Miller refers to similarities that

lead to essential rather than accidental consensus (in Fuller’s (2002) terms). However, he does

not positively justify why formalism, ontological schemes and evidential standards have to be

shared, but why nothing more has to, pace Fuller’s claim that the scientists’ interpretation and

conceptual understanding of a theory also need to be shared. For Miller, a consensus can

benefit from a diversity of interpretations and understandings of a theory. This apparent claim

for uniformity boils down to another defence of background diversity, which virtues would

equally guarantee. Overall, scientific virtues preserve the effect of Miller’s conditions, thus

favouring successful consensus.

We are not claiming that scientific virtues are sufficient to lead to consensus though.

Since Kuhn (1977), it is common knowledge that scientists can legitimately disagree if they

favour different theoretical virtues (theory properties such as simplicity, fruitfulness, etc.).

Virtues do not, and should not, eliminate such dissent.33 Moreover, virtues need not be co-

instantiated. Scientists who exemplify different virtues may not agree, because they are still

influenced by remaining non evidence-based reasons. Our claim is that, other things being

equal, the presence of virtues in a community favours successful consensus.

The foregoing leads us to a useful analogy between virtues and catalysts. In chemistry,

catalysts are substances that accelerate specific reactions while staying unchanged. When in

presence of the right starting elements, they speed up the appearance of the products. The role

of scientific virtues with respect to scientific success is akin to that of catalysts in two respects.

First, as argued above, scientific virtues favour convergence on a theory. The more virtuous

scientists are around, the faster a critical number may converge on a good theory. Another

way to put it is to say that as virtues make scientists more sensitive to evidence (relatively to

other factors that influence theory choice) they make it more likely that there will be

convergence on a theory.34 Second, this convergence can only lead to (at least temporary)

success if a good enough theory is available to begin with: virtues only work in presence of

the right ingredients.35 Briefly put, scientific virtues are akin to catalysts because they increase

the odds of successful convergence in presence of the right ingredients.36

Again, this does not make virtues sufficient for scientific success. Virtues only

manifest their speeding-up effects in presence of good enough theories; and vices can be very

efficient in producing diverse theories and widening the list of options. Theories must be

33 Note that this strays from Duhem (1954), who thought that a decisive consensus always appears to favour the

best choice in the end. However, as highlighted by Solomon and Miller, consensus can be bad, and dissent good. 34 This does not damage the analogy, as catalysts often boost a reaction rate by increasing the probability that a

reaction will happen - for instance by lowering its activation energy. 35 Incidentally, one could push the analogy by stressing a third analogous point: just as catalysts are unaffected

by the chemical reaction, virtues, as character traits, are not affected by the convergence process. However,

although true, this remark adds nothing to our understanding of theory choice. 36 Recall that by ‘successful’ we do not mean that the result should be a true or even a perfectly empirically

adequate theory. Success could be limited and temporary. In other words, we do not claim that virtues guide us

towards truth.

14

created and explored; as a consequence, overall scientific efficiency benefits from both vices

and virtues.

8. Discussion

Our analysis of the role of intellectual virtues might come under criticism from existing

accounts, according to which scientific success is not due to properties of the scientists but of

scientific communities: what ultimately accounts for scientific success is the social structure

of scientific inquiry. It may also be deemed too similar to other existing accounts.

Merton (1973) famously emphasised the structural features of science over the role of

personal virtues. According to Merton, science produces knowledge despite the subjective

values of the individual participants because of the communal norms responsible for the

correction of individual behaviours. There are four such ‘institutional imperatives’ (1968, p.

608): (1) universalism; (2) communism; (3) disinterestedness, not on the individual but on the

institutional level; and (4) organized scepticism. In effect, two of these are equivalent to the

sharing of individual virtues. Universalism imposes that scientific statements be assessed

through ‘preestablished, impersonal criteria’. Similarly, organized scepticism involves ‘the

suspension of judgment until 'the facts are at hand' and the detached scrutiny of beliefs in

terms of empirical and logical criteria’. These amounts to all scientists being impartial at least,

intellectually sober (in the broad sense) at most. As for communism, it states that the products

of science are collective and belong to the whole community, and as such is neither about

individual nor about structural properties. Only disinterestedness poses a challenge, as it

concerns structural features supposed to neutralise the effects of individual ideological biases.

Longino (1990) also claims that scientific knowledge is objective despite the presence

of personal values. According to Longino’s account, despite the fact ‘contextual values’

constitute the construction and evaluation of scientific theories, the objectivity of scientific

inquiry is not undermined because science is a group activity that allows for ‘transformative

criticism’, which limits the ill-founded contextual values. Longino provides four necessary

criteria for scientific objectivity: (1) ‘recognised avenues for criticism’, an organised way to

communicate criticism (e.g. journals and conferences); (2) ‘shared standards’ or values that

all critics invoke; (3) ‘community responsiveness’, or readiness to revise the community’s

beliefs; and (4) ‘equality of intellectual authority’: the intellectual authority must be equally

distributed among practising scientists.

If Merton’s disinterestedness and Longino’s criteria contribute to collective efficiency

in science, what role do individual virtues play? We saw that virtues are neither sufficient nor

necessary for scientific success; they merely make it more likely. It may be that scientific

success would be extremely unlikely without adequate structural features, to which virtues

may then add very little. Our aim, however, is not to analyse how and why science is

successful, but to understand how individual virtues affect this success. Intellectual and moral

virtues can favour successful convergence while other factors are at play - even if they are

necessary and sufficient for scientific success. There is no conflict between the structural

claims and ours. Indeed, the adequate features that make scientific convergence possible in

the first place are probably structural.

Still, our account of scientific virtues is opposed to any account that makes scientific

virtues reducible or eliminable to others. McMullin notes that it is broadly accepted that

ethical values, such as honesty, openness and integrity, are “essential to the success of

communal inquiry” (1983, 9). Other ‘non-epistemic’ values such as personal interests,

ambition, social class, political, social or religious memberships, etc., are eliminated through

scientific progress and can at best be regarded as instrumental. On this point we would follow

15

Ruse (1984), who argues that non-epistemic values should not - and in any case cannot - be

eliminated from scientific practice, as they are inherently present in the choice of research

topic, language and are reflected in the very theories that are constructed by the scientific

community. Still, we emphasise that vices too would be useful in the pre-convergence phase

of scientific activity.

Finally, our account may seem similar to Solomon’s (2001) social empiricism.

Solomon contends that consensus, because it may be attained for non-empirical reasons, ‘is

not in itself of mush epistemic significance. The pursuit of empirical success and of truth can

be consistent with both dissent and consensus’ (2001, 11). According to Solomon, consensus

can result from many factors (‘decision vectors’ in her terms), either empirical or non-

empirical. Importantly, consensus on the empirically preferable theory may emerge from non-

empirical reasons. This can be interpreted as a precursor to our claim that vices have

collective epistemic benefits. However, we make no claim about the general positive or

negative effects of consensus or dissent, nor do we discuss the frequency of their occurrences

in science. We claim that virtuous scientists are more likely to participate in a successful

consensus. This is because they tend to focus on empirical factors and disregard non-

empirical ones. In Solomon’s terms, the less numerous non-empirical decision factors are, the

less likely the resulting consensus is to be ‘bad’. As noted above, virtue-based consensus may

be partial, which allows for a state of general dissent in the whole scientific community (as

Solomon argues should be the most frequent case). Our accounts are thus not competing or

overlapping but compatible and complementary.

Let us go back to our initial question again: does science benefit from virtues?

Existing diversity models suggested that vices can be beneficial to science; we later showed

that anything goes in pre-convergence science – and vices in particular fuel theory diversity.

Even in post-convergence science, once research communities have formed, vices can be

tolerated and even necessary, a familiar lesson since Kuhn. What room is left for virtues? The

answer is that scientists who share similar virtues are more likely to favour the same

hypotheses or theories and so to gather in order to work on them. Virtuous scientists are most

likely to aggregate so as to form successful research communities. Dogmatic and biased

scientists may be more inspired when creating, promoting a theory and refining it to survive

apparently conflicting evidence; but not when assessing its advantages with respect to rival

theories. More importantly, dogmatic scientists will not tend to converge on what is most

promising as much as evidence-sensitive ones. Research groups are more likely to be

successful when comprised of, or originating from, virtuous scientists. To put it differently,

successful groups will tend to be composed of a higher number of virtuous scientists, even if

this tendency may only be clear when considering all possible worlds.

In practice, virtues should not be expected to be exemplified in any research group:

even if there is ever such a thing as an objectively preferable theory, virtuous scientists who

focus on it would join those who were already biased towards it (for whatever reason).

Moreover, even if virtues, which are science-relevant personal characteristics, were

exemplified in a research group, scientists may still perform different activities. Finally, this

account does not entail that we should strive to make scientific communities more virtuous as

a whole, as this may severely cripple efficient theory generation.

9. Conclusion

In general, scientific virtues are not crucial for scientific success. Science cannot be successful

if promising enough hypotheses or theories do not appear; vices (and other factors) bolster

this initial diversity of theories. Once research groups have gathered to refine a theory, vices

16

that make scientists cling to it can also be beneficial, as they may postpone theory rejection

and leave time for a successful refinement to appear.

Intellectual virtues can also foster scientific success, although in a different way.

Similarly virtuous scientists are more likely to make similar theory choices that make them

converge on promising theories. Scientific virtues act like catalysts, with which they share

two key characteristics. They accelerate the scientific process by favouring the development

of successful theories; they do that only when the promising theories are already available.

Scientific virtues are neither necessary nor sufficient for scientific success. Rather, they

increase the odds of successful convergence in the presence of the right ingredients.

References

Allchin, D. (1999) Values in Science: An Educational Perspective. Science and Education, 8,

1-12.

Axtell, G. (2014) Bridging a Fault Line: On Underdetermination and the Ampliative Adequacy

of Competing Theories, in Virtue Epistemology Naturalized (A. Fairweather ed.),

Synthese Library, 366, 227-245.

Baehr, J. (2011). The Inquiring Mind: On Intellectual Virtues and Virtue Epistemology. NY:

Oxford University Press.

Daston, L.J. and Gallison, P. (2007). Objectivity. M.I.T. Press.

De Langhe, R. (2010). The Division of Labour in Science: the Tradeoff between

specialisation and diversity. Journal of Economic Methodology 17(1), 37-51.

Di Bucchianico, M. (2014). A Matter of Phronesis: Experiment and Virtue in Physics, A Case

Study, in Virtue Epistemology Naturalized, A. Fairweather (ed.), Synthese Library,

366, 291-312.

Duhem, P. (1954 [1906]). The Aim and Structure of Physical Theory. Princeton: Princeton

University Press.

Duhem, P. (1991 [1915]). German science: Some reflections on German science: German

science and German virtues (John Lyon, Trans.). La Salle, IL: Open Court.

Fairweather, A. (2011). The Epistemic Value of Good Sense. Studies in the History and

Philosophy of Science, 43(1), 139-146.

Feyerabend, P. (1970). Consolations for the Specialist in Criticism and the Growth of

Knowledge, I. Lakatos and A. Musgrave (eds), 197-230. Cambridge: Cambridge

University Press.

Franklin, A. (1998). Selectivity and the Production of Experimental Results. Archive for

History of Exact Sciences, 53(5), 399-485.

Fricker, M. (2012). Group Testimony? The Making of a Collective Good Informant.

Philosophy and Phenomenological Research 54(2), 249-276.

Fuller, S. (2002). Social Epistemology (2nd ed.) Bloomington, IN: Indiana University Press.

Geison, G.L. (1995). The Private Science of Louis Pasteur. Princeton University Press.

Gettier, E. (1963). Is Justified True Belief Knowledge? Analysis, 23, 121–123.

Greco, J. (2002). Virtues in Epistemology. In The Oxford Handbook of Epistemology, Paul

Moser (ed.), 287-315. Oxford: Oxford University Press.

Hull, D. (1988). Science as a Process: An Evolutionary Account of the Social and

Conceptual Development of Science. The University of Chicago Press.

Kawall, J. (2002). Other-Regarding Epistemic Virtues. Ratio, 15(3), 257-275.

Kidd, I.J. (2011). Pierre Duhem’s epistemic aims and the intellectual virtue of humility:

a reply to Ivanova. Studies in the History and Philosophy of Science Part A, 42(1),

185-189.

17

Kidd, I.J. (2014). Was Sir William Crookes Epistemically Virtuous? Studies in the

History and Philosophy of Biological and Biomedical Sciences, 48, 67-74.

Kitcher, P. (1990). The Division of Cognitive Labor. Philosophy of Science, 87, 5-22.

Kuhn, T. (1977) Objectivity, Value Judgment, and Theory Choice. In The Essential Tension

(pp. 320-353). The University of Chicago Press.

Longino, H. (1990) Science as Social Knowledge: Value and Objectivity in Scientific Inquiry.

Princeton: Princeton University Press.

McMullin, E. (1983). Values in Science. In PSA 1982, Vol. 2. Edited by P.D. Asquith and T.

Nickles. East Lansing, MI: Philosophy of Science Association, 3-28.

Merton, R. (1957) Priorities in Scientific Discovery. American Sociological Review, 22, 639-

659.

Merton, R. (1973). The Sociology of Science: Theoretical and Empirical Investigations.

University of Chicago Press.

Miller, B. (2013). When is Consensus Knowledge Based? Distinguishing Shared Knowledge

from Mere Agreement. Synthese, 190, 1293-1316.

Morton, A. (2014a). Acting to Know: A Virtue of Experimentation, in Virtue Epistemology

Naturalized (A. Fairweather ed.), Synthese Library, 366, 195-205.

Morton, A. (2014b). Shared Knowledge from Individual Vice: the Role of Unworthy

Epistemic Emotions. Philosophical Inquiries 2(1), 163-172.

Montmarquet, J., (1993). Epistemic Virtue and Doxastic Responsibility. Lanham: Rowman

and Littlefield.

Muldoon, R. and Weisberg, M. (2011). Robustness and Idealization in Models of Cognitive

Labor. Synthese, 183(2), 161-174.

Okruhlik K. (1994). Gender and the Biological Sciences. Canadian Journal of Philosophy,

20, 21-42.

Putnam, H. (1981). Reason, Truth and History. Cambridge: Cambridge University Press.

Roberts, R.C. and Wood, W.J. (2007). Intellectual virtues: An Essay in Regulative

Epistemology. Oxford, NY: Clarendon Press.

Rowbottom, D. (2011). Kuhn vs. Popper on Criticism and Dogmatism in Science: a

Resolution at the Group Level. Studies in History and Philosophy of Science, 42,

117-124.

Ruse, M. (1984). Biological Science and Feminist Values. In PSA 1984, Vol. 2. Edited by

P.D. Asquith and P. Kitcher. East Lansing: Philosophy of Science Association. 525-

542.

Shapin, S. (2008). The Scientific Life: A Moral History of a Late Modern Vocation. The

University of Chicago Press.

Solomon, M. (2001). Social Empiricism. Cambridge, MA: MIT Press.

Stanford, P.K. (forthcoming). Unconceived Alternatives and Conservatism in Science :

the impact of professionalization, peer-review, and Big Science. Synthese. DOI

10.1007/s11229-015-0856-4

Strevens, M. (2003). The Role of the Priority Rule in Science. Journal of Philosophy, 100(2),

55-79.

Stump, D. (2007). Pierre Duhem’s Virtue Epistemology. Studies in History and Philosophy of

Science, 38, 149-159.

Tsou, J.Y. (2003). Reconsidering Feyerabend’s “Anarchism”. Perspectives On Science, 11(2),

208-235.

Tucker, A. (2003). The Epistemic Significance of Consensus. Inquiry, 46, 501-521.

Vallor, S (2014). Experimental Virtue: Perceptual Responsiveness and the Praxis of Scientific

Observation, in Virtue Epistemology Naturalized (A. Fairweather ed.), Synthese

Library, 366, 269-290.

18

Weisberg M. and Muldoon R. (2009) Epistemic Landscapes and the Division of Labour.

Philosophy of Science, 76, 225-252.

Zagzebski, L. (1996). Virtues of the Mind. Cambridge: Cambridge University Press.

Zagzebski, L. (2003). The Search for the Source of Epistemic Good. Metaphilosophy, 34, 12-

28.

Zollman, K.J.S. (2010). The Epistemic Benefit of Transient Diversity. Erkenntnis, 72, 17-35.