Synthese

DOI 10.1007/s11229-013-0383-0

Knowledge and cognitive integration

Spyridon Orestis Palermos

Received: 10 July 2013 / Accepted: 4 December 2013

© Springer Science+Business Media Dordrecht 2013

Abstract Cognitive integration is a defining yet overlooked feature of our intellect that
may nevertheless have substantial effects on the process of knowledge-acquisition. To
bring those effects to the fore, I explore the topic of cognitive integration both from the
perspective of virtue reliabilism within externalist epistemology and the perspective
of extended cognition within externalist philosophy of mind and cognitive science. On
the basis of this interdisciplinary focus, I argue that cognitive integration can provide
a minimalist yet adequate epistemic norm of subjective justification: so long as the
agent’s belief-forming process has been integrated in his cognitive character, the agent
can be justified in holding the resulting beliefs merely by lacking any doubts there
was something wrong in the way he arrived at them. Moreover, since both externalist
philosophy of mind and externalist epistemology treat the process of cognitive inte-
gration in the same way, we can claim that epistemic cognitive characters may extend
beyond our organismic cognitive capacities to the artifacts we employ or even to other
agents we interact with. This move is not only necessary for accounting for advanced
cases of knowledge that is the product of the operation of epistemic artifacts or the
interactive activity of research teams, but it can further lead to interesting ramifications
both for social epistemology and philosophy of science.

Keywords Knowledge · Cognitive integration · Subjective justification · Virtue

reliabilism · Cognitive character · Extended cognition · Epistemic group agency

1 Introduction

Cognitive integration is an overlooked yet defining feature of our intellect. It is the

reason why, in contrast to mere stimulus–response automata, we entertain advanced

S. O. Palermos (B)
Extended Knowledge Project, University of Edinburgh, Edinburgh, UK
e-mail: [email protected]

123
Synthese

beliefs in an epistemically responsible way, and why we can do so even in the complete
absence of any reasons to back those beliefs up.
Nevertheless, within philosophy of mind and cognitive science, it is only recently
that the topic of cognitive integration was brought into focus. This recent change
of focus, however, has not been and is still not guided by an attempt to understand
how our complex brain capacities intertwine with each other. Due to Fodor’s (1983)
persuasive understanding of our intellectual architecture as modular, in-the-head cog-
nition has largely thought to be for the most part domain specific and informationally
encapsulated.1 Consequently, cognitive scientists who take the brain to be the primary
cognitive explanandum have so far been indisposed to explore and understand the
phenomenon of cognitive integration.
Instead, only after the hypothesis of extended cognition (Clark and Chalmers 1998)
was proposed (according to which external elements can be proper parts of our cogni-
tive systems), did the topic of cognitive integration start to attract attention; claiming
that artifacts can be parts of an agent’s cognitive system presupposes an account of how
such external elements can be properly integrated into our cognitive loops. Accord-
ingly, over the past fifteen years, several attempts have been made to account for the
process of cognitive integration, most of which rely either on a common-sense func-
tionalist understanding of our minds (Clark and Chalmers 1998; Clark and Chalmers
2010), or on a mathematically inspired approach (Clark and Chalmers 2008; Chemero
2009; Palermos 2014) that focuses on the dynamical nature of cognitive processes.
Interestingly, however, the topic of cognitive integration has also recently emerged
within epistemology in reference to the concept of our epistemic cognitive charac-
ters. Even within epistemology, however, the discussion of cognitive integration has
not been entirely unrelated to the idea of cognitive extension. Admittedly, Greco
(2010)—the first to discuss (in passing) the idea of cognitive integration in epistemo-
logical terms—does not commit himself to the possibility of cognitive extension. Both
Pritchard (2010b) and I Palermos (2011), however, have noted that the notion of the
epistemic agent’s cognitive character, to which all of the agent’s knowledge-conducive
belief-forming processes must have been properly integrated, is open to an interpre-
tation along the lines suggested by the extended cognition hypothesis. However, and
despite these few efforts to elucidate the process of cognitive integration within epis-
temology, more needs to be said since the effects of this process on knowledge could
turn out to be surprisingly substantial.
To bring these effects to the fore, we need to approach the idea of cognitive integra-
tion both from the perspective of virtue reliabilism in externalist epistemology (Sect.
2.2) and the perspective of extended cognition in externalist philosophy of mind (Sect.
2.3). This interdisciplinary focus will help us make apparent, in Sect. 2.3, that both
views treat cognitive integration in essentially the same way—an observation that we
can then use to bring both perspectives together and thereby demonstrate how our
epistemic cognitive characters can extend beyond our organismic cognitive capacities
to the epistemic artifacts we employ. This combination of externalist epistemology

1 Despite the influence of Fodor’s work within philosophy of mind and cognitive science, his modular
understanding of the mind has met some considerable resistance by equally influential philosophers. One
clear example is the work of Churchland (1979, 1988, 1989).

123
Synthese

with externalist philosophy of mind, however, as I will argue in Sect. 3, is not just
an available option, but is actually necessary for accounting for advanced cases of
knowledge whereby one’s true believing is the product of the operation of epistemic
artifacts. Furthermore, this approach can generate interesting ramifications both for
social epistemology and philosophy of science, wherein the pursuit of knowledge
has been traditionally associated with the use of artifacts, carefully tailored labs, and
the combined efforts of epistemic agents working in research teams. To start with,
however, a few introductory remarks are first in order.

2 The process of cognitive integration

2.1 Introduction

“John is sad” is an assertion that can be true or false. In producing this claim, of course,
perception will always be foundational in a certain way, but in order to make, as well
as check, the validity of the claim we combine several other processes as well. In the
background of our minds we have a biological as well as a primordial (animalistic)
conception of what a human being is and we also have a socio-contextual sense of who
“John” is; we have a theory of mind that enables us to understand other people with
thoughts and emotions like ours, and we may even need a more personal theory of
John’s character: John has spent the entire night socializing at the pub and he just made
a witty joke, but the downwards motion of his eyes accompanied by a melancholic
smile at the end of his remark are enough to indicate to his close friends that his mind
is secretly occupied with his recent loss.
It is in a sense like this that theories (commonsensical as well as scientific ones) have
a top-down effect on the bottom-up input we receive from our sense-apparatuses. This
overriding effect, known as the theory-ladeness of observations, has been well-studied
both within philosophy of science (Kuhn 1962; Hanson 1961, 1969) and philosophy of
mind (Churchland 1979, 1988, 1989; Fodor 1984, 1988), and indicates that perceptual
beliefs are never ‘purely perceptual’, but they instead emerge out of the cooperation
of several intellectual capacities operating in tandem.
Now, considerations like these indicate that perception (traditionally thought of as
the most foundational aspect of our belief systems) may not be all that basic, and
have thereby hinted towards a relativistic picture of science—and possibly epistemol-
ogy as well—whose most prominent proponents are thought to be Feyerabend (1975)
and Kuhn (1962) (the latter quite possibly unjustly though). Fortunately, however, at
least as far as the theory-ladeness of observation is concerned, recent studies within
cognitive psychology not only point away from relativism, but they also demonstrate
how the interaction between our theoretical beliefs and our sensory apparatus has a
facilitatory effect on the overall process of perception. For example, exploring the
literature on relevant experiments, Brewer and Lambert (2001) concede that “percep-
tion is determined by the interaction of top-down theory information and bottom-up
sensory information” (p. 178, my emphasis):
However, note that in all of the above cases the stimuli were either ambiguous,
degraded, or required a difficult perceptual judgment. In these cases the weak

123
Synthese

bottom-up information allowed the top-down influences to have a strong impact

on perceptual experience. It seems likely that strong bottom-up information
will override top-down information. [...] Thus, the topdown/ bottom-up analysis
allows one to have cases of theory-laden perception, but does not necessarily
lead down the slippery slope of relativism (ibid.). 2
Therefore, we should neither be misguided nor feel intimidated by the possibly false
impression of relativistic bias that the integration of our intellectual capacities into
one overarching belief-generating system may have initially created. To the contrary,
we should focus on the phenomenon of cognitive integration because, as we shall see,
it actually has a very important facilitatory effect on our epistemic standing. In par-
ticular, the interconnectedness of our cognitive capacities allows us to be subjectively
justified even if we lack explicit reasons for holding our beliefs. Provided that we act
in a conscientious mode—i.e., provided we are motivated to believe what is true—
cognitive integration allows us to trust the deliverances of our cognitive abilities by
merely lacking any negative reasons against (as opposed to possessing positive reasons
for) our beliefs. This sense of epistemically adequate—yet unreflective—cognitive
responsibility can only be achieved by agents like us, whose intellectual capacities
are appropriately interconnected such that in cases where there is something wrong
with the way we form our beliefs or with the beliefs themselves, we will be able to
notice this and respond appropriately. Otherwise—if there is nothing wrong—we can
go on about with our daily activities without questioning our epistemic standing with
respect to every single of the millions (possibly billions?) of beliefs we enjoy in the
course of our days.
On a first pass, this probably sounds sketchy, but focusing on contemporary epis-
temology should allow us to both understand what this unreflective sense of cognitive
responsibility amounts to and why it is so important.

2.2 Cognitive integration in epistemology

To start with, consider epistemic internalism, which takes an approach to epistemic

responsibility that is very different from the one suggested here. According to tradi-
tional forms of internalism, one should always be able, at least in principle, to access
the reasons that justify one’s beliefs, by reflection alone.3 This may initially sound

2 Similarly, (Estany (2001), p. 208) holds that

the beliefs of the higher or more fundamental level influence how perceptual units are interpreted
by the lower levels [...] Humans use both types of processes in perception because each have char-
acteristic advantages and disadvantages. Thanks to top-down processes we can recognize patterns
with incomplete or degraded information. Moreover, top-down processes make perception faster,
but they can induce us to make mistakes in a perception by relying on previous knowledge.
Nevertheless, Estany further notes that even though our perceptual systems get guidance from higher-order
expectations, when attention is caused by the mismatches between expectation and reality, the inputs from
the arousal system constitute a “reset wave” making it possible to avoid arbitrary relativistic errors of
perception (Estany 2001, p. 213).
3 For classical defenses of this view see Chisholm (1977) and BonJour (1985, Chap. 2). See also Steup
(1999), Pryor (2001, p. 3), BonJour (2002), Pappas (2005), and Poston (2008).

123
Synthese

as a reasonable demand, but the problem is that it creates serious complications with
respect to our perceptual and empirical beliefs. Specifically, it poses the requirement
that there be necessary support relations between one’s empirical and perceptual beliefs
and one’s evidence for holding them (such that one can be in a position to justify one’s
empirical and perceptual beliefs by reflection alone). As Hume’s problem of induc-
tion demonstrates, however, this is impossible. Accordingly, it has been traditionally
assumed that Hume’s arguments lead to skepticism about our empirical knowledge.4
Contemplating on the Humean problematic, however, Greco (1999) argues that
this is too fast. Hume’s arguments should not be directly considered as skeptical
ones. Instead, the immediate conclusion to be drawn from them is that there are no
necessary support relations between our empirical beliefs and their evidence; that
if the evidence for our empirical beliefs is reliable, then it is at most contingently
reliable. This realization alone, however, cannot automatically lead to skepticism.
Only after we embrace the internalist understanding of knowledge, such that there
always be necessary support relations between one’s evidence and one’s beliefs do we
face skepticism.
In other words, in order to avoid skepticism about empirical and perceptual knowl-
edge, we must allow knowledge to be grounded on evidence that is merely contingently
reliable, and so we must give up the requirement that one’s beliefs should always be
internally—i.e., by reflection alone—justified. Any adequate epistemology must be
able to account for the fact that merely contingently reliable evidence can give rise to
knowledge (Greco 1999, p. 273).
Now, in order to accommodate the above realization, contemporary epistemologists
have put forward process reliabilism; viz., the idea that knowledge is true belief that is
the product of reliable belief-forming processes, where a reliable process is a process
that results in a preponderance of true over false beliefs. Moreover, in direct response
to the Humean problematic, this approach “denies that one must know that one’s
evidence is reliable”, by making “de facto reliability the grounds of positive epistemic
status” (Greco 1999, pp. 284–285). Accordingly, process reliabilism is an externalist
approach to knowledge, because—contrary to the traditional account of knowledge
as internally justified true belief—on this view, in order to know, one does not need
to know, be justified in believing (by reflection alone, or any other means), or even
believe that one’s beliefs are formed in a reliable fashion. So long as one employs an
objectively reliable process, one is justified in holding the resulting belief.
Process reliabilism, therefore, has the resources to overcome the Humean skepti-
cism. There are, however, two serious complications with the view. The first one is

4 The problem of induction is well known. We form our beliefs about unobserved matters of fact and the
external world on the basis of evidence provided by past and present observations and sensory appearances,
respectively. In order, however, for the support relations between our empirical and perceptual beliefs and
the evidence offered in their support to be necessary, we also need the further assumptions that the future will
resemble the past and that sensory appearances are reliable indications to reality, respectively. The problem,
however, is that both of these assumptions rely for their support on what they assert. Consequently, given
that circular reasoning is invalid, there are no necessary support relations between our empirical beliefs and
the evidence offered in their support. Accordingly, the conclusion that has been traditionally drawn is that
our empirical and perceptual beliefs cannot amount to knowledge. For more details on a reconstruction of
Hume’s skepticism along these lines, see (Greco 1999).

123
Synthese

that process reliabilism, as it stands, is too weak a condition on knowledge because it

allows any reliable belief-forming process to count as knowledge-conducive, and this,
as we shall see, is intuitively incorrect. The second complication is that by making “
de facto reliability the grounds of positive epistemic status” process reliabilism misses
a very important dimension of our epistemic nature. While it is true that in order to
know we do need the way of forming our beliefs to be objectively reliable, this sort
of objective justification is not sufficient in its own. What we further need is that we
be subjectively justified in the sense that we must be somehow sensitive to the reli-
ability of our evidence.5 Process reliabilism, however, ignores this dimension of our
epistemically sentient nature altogether, to the extent that it has been even criticized
that it equates us to mere stimulus–response automata (Fuller 2012).6 We can better
appreciate these two problems by taking a look at a few (eccentric, yet informative)
examples. Consider Hercules first:
Hercules [Adapted from Pritchard’s Temp (2009, p. 48)]
Hercules tosses a drachma whenever he wants to form a belief about the weather
outside. If it is heads, he forms the belief that it is sunny; if it is tails he believes
it is cloudy; and if it balances in between, he believes it is rainy. As it happens,
Hercules’ way of forming his weather beliefs is perfectly reliable, because Zeus,
who wants to save Hercules from the embarrassment of forming false weather
beliefs, has an eye on him; every time he sees Hercules tossing the coin arranges
the world accordingly.

Hercules’ beliefs are formed in a highly reliable way. So, according to process relia-
bilism, Hercules has knowledge of the weather conditions. Intuitively, however, this is
incorrect. There is a problem with the direction of fit between his beliefs and the facts.
In cases of knowledge, we want our beliefs to be true because they correspond to the
facts, and not because the facts comply with our beliefs; when one knows, one’s true
beliefs are about the world, not the other way around. In Hercules’ case, however, his
beliefs are not true because they are formed in a way that detects the facts. Instead,
he first forms his beliefs in an arbitrary way—he makes no efforts to ensure they will
come out true—and then Zeus takes over so that the facts will comply with Hercules’
beliefs. This, however, is not knowledge; it is the ‘luck of the gods’. If one day Zeus
had a fight with Hera, Hercules’ beliefs would cease coming out true.
Notice, however, that if Hercules used his cognitive abilities—say by taking a look
at the sky—to form his weather beliefs, then he would not run into any such problems.
If he didn’t form his beliefs in an arbitrary way, but on the basis of his cognitive abilities,
he would not need Zeus to tweak the world so that his beliefs could systematically
turn out true. If one’s beliefs are the product of one’s cognitive abilities, then if they
5 Remember, however, that if, as Hume’s skeptical arguments demonstrate, the relation between evidence
and belief is not necessary (see also fn. 4), then it is far from obvious how a person can be subjectively
justified, especially in externalist approaches such as process reliabilism. If a condition of ‘subjective
sensitivity to the reliability of one’s evidence’ must be satisfied, then this should better be accomplished
in a way that will not require knowledge of or even beliefs about the said reliability (otherwise Hume’s
skepticism will strike back).
6 “Epistemic zombies” would probably be the name that David Chalmers would give to such creatures.
Given, however, the present discussion I don’t think they could really exist.

123
Synthese

turn out to be true it will be because they are sensitive to the facts; the direction of fit
will be the correct one.
So, it may be proposed that the way to restrict the reliable belief-forming processes
to those that get the direction of fit correctly—such that they can be knowledge-
conducive—is to identify them with one’s cognitive abilities, or, in other words, with
those processes that can be intuitively thought of as cognitive ones. But can all prima
facie cognitive processes count as cognitive abilities and thereby produce knowledge?
The answer, as we shall now see, must be a negative one, because there are reliable
processes that we might be inclined to categorize as cognitive ones, but which fail to
deliver knowledge, exactly because they disallow the agent to be subjectively justified
in employing them. Think about the Serendipitous Brain Lesion first:
Serendipitous Brain Lesion (Greco 2010, p. 149)
Suppose that S has a rare brain lesion, one effect of which is to reliably cause the
true belief that one has a brain lesion. Even if the process is perfectly reliable, it
seems wrong that one can come to have knowledge that one has a brain lesion
on this basis.
Again, the unfortunate agent’s way of forming his true belief about the brain lesion on
the basis of his brain lesion is reliable. Process reliabilists, therefore, must accept that
he can gain knowledge in this way. As Greco claims, however, this does not sound
correct. Why not? Mainly because the way the agent forms his belief is so strange
from his point of view that he cannot accept he can gain knowledge in this way (Greco
1999, 2010). Accordingly, there might be reliable in-the-head processes (such that we
may be inclined to call them cognitive ones) that we wouldn’t like to claim they are
knowledge-conducive cognitive abilities, because from the agent’s point of view they
are strange. More precisely, the underlying intuition here is that for a process to be
eligible to count as a cognitive ability it must not be strange, in the sense that it must
not be at odds with the rest of the agent’s cognitive system.7 The reason is that if the
process is strange, then, in light of the rest of his cognitive system, the agent will reject
both the process and its deliverances despite the fact that they are in fact reliable—from
the agent’s point of view, they aren’t. So, in order for a process to be a candidate for
qualifying as a cognitive ability such that it can be knowledge-conducive it must not
be inconsistent with the rest of the agent’s beliefs and his methods of producing them.
In other words, it must be such that it can become part of, or be integrated into, the
rest of the agent’s cognitive system (Greco 2010, p. 152). And clearly, this is not the

7 An anonymous referee points out that strangeness is description-relative. Take vision for example. We
are all familiar with acquiring knowledge through seeing things. But learning about the physiological and
neural underpinnings of vision will surely seem strange to some; couldn’t such a person say “This is really
strange, and I don’t really see how it works, but, I guess, this is how I know the color of my shoes”? I think
this is right, but this example wouldn’t be problematic for the following two reasons. First, even though
the explanation may seem ‘strange’ to the agent (in the sense of being difficult to understand) it is not ‘at
odds with the rest of the agent’s cognitive system’. Second, the requirement that the process not be strange
does not refer to a reflective-explanatory understanding of the process (as in the referee’s example), but
to the presence of the process itself. Think about the analogy of a strange (i.e., eccentric) person who is
nevertheless not a stranger: the requirement that the process not be strange allows for the process to be
‘strange’ in the first sense, but not in the sense of being a ‘stranger’. I am thankful to the referee for bringing
this ambiguity to my attention.

123
Synthese

case with the serendipitous brain lesion. The process is a cognitive malfunction, and
even more crucially, its output is so odd that no epistemic agent could accept as true.
In other words, the serendipitous brain lesion cannot count as knowledge-conducive
because it is so strange that it cannot become part of the rest of the agent’s cognitive
system and so cannot count as a cognitive ability.8
Nevertheless, even a reliable process that is normal enough to become part of
the agent’s cognitive system cannot yet count as a cognitive ability that can produce
knowledge. Consider a further example:
Careless Math Student (Greco 2010, p. 149)
Suppose that S is taking a math test and adopts a correct algorithm for solving
a problem. But suppose that S has no understanding that the algorithm is the
correct one to use for this problem. Rather, S chooses it on a whim, but could
just as well have chosen one that is incorrect. By hypothesis, the algorithm is the
right one, and so using it to solve the problem constitutes a reliable process. It
seems wrong to say that S thereby knows the answer to the problem, however.
The careless student’s algorithm for solving the problem is also reliable. But again, we
cannot attribute knowledge to her. Why not? The reason is that she employed the right
method on a whim, such that she could have very easily employed another, incorrect
method. Her reliable process is a fleeting one. It is not a habit or a disposition of hers.
Given the same circumstances, she could have employed an inappropriate method,
thereby, ending up with a falsehood. If, however, the student had habitually invoked
the correct algorithm when the problems called for it, then we would indeed be inclined
to claim that she could gain knowledge on its basis. The reason for this is that if a
process is a disposition or a habit of the agent, then the agent will be able to become
aware of the circumstances in which it can become unreliable. Otherwise, it seems
arbitrary that the agent employed it in an appropriate, but isolated case, and so cannot
gain knowledge on its basis. In other words, a reliable process that is normal—such
that it can, in principle, become part of the rest of one’s cognitive system—won’t
be a candidate for qualifying as a cognitive ability, unless it is also a disposition or
a habit of the agent. Why is this so? The intuition is that abilities, in general, are

8 An anonymous referee insists that the brain lesion can yield knowledge, thereby implying that process
reliabilism is a sufficient condition on knowledge. While it may be possible to make the case for the
sufficiency of process reliabilism, the orthodox view within mainstream epistemology goes against this
prospect. For classical rejections of the sufficiency of process reliabilism on the basis of thought experiments
very similar to the Serendipitous Brain Lesion, see BonJour (1980), Lehrer (1990) and Plantinga (1993).
The main idea is that reliability might be necessary for knowledge but what is further required is satisfaction
of the internalist intuitions with respect to the possession of subjective justification (as we mentioned above,
one of the problems for process reliabilism is that by making de facto reliability the grounds of positive
epistemic status, it fails to capture the intuition that, somehow, we must also be sensitive to the reliability
of our evidence). Internalists, typically require the possession of reflectively accessible reasons for said
reliability. Here, following Greco’s intuitions (1999, 2010, pp. 149–155), we opt for a weaker condition of
subjective justification according to which the agent must lack beliefs against the reliability of his belief-
forming process, where the process being strange from the agent’s point of view would count as just one such
defeating belief. For very similar intuitions on how the strangeness of the origin of the relevant beliefs acts
as a defeater in BonJour (1980) and Lehrer (1990) thought experiments see Goldman (1986, pp. 111–112).

123
Synthese

habits or dispositions possessed by agents.9 But apart from such intuitions we have
also noted that in order for a reliable process to count as a cognitive ability it must be
such that it can become part of (or be integrated into) the rest of the agent’s cognitive
system. One requirement for this, we have noted, is that the process not be strange
such that it won’t be inconsistent with the rest of the agent’s cognitive system. What
is further required, however, is that it also be coherent with her cognitive system in
the following sense: The agent must be able to become aware that the process is
unreliable in certain circumstances, because this will allow her to non-accidentally
endorse its deliverances in the rest of the circumstances, even if she lacks any positive
beliefs for its reliability. And in the absence of any explicit reasons that are accessible
through reflection alone—recall the Humean problematic—the only realistic way for
the agent to so become epistemically responsible in employing a process is that it be
a disposition or habit of hers.10,11
So to summarize what we have gathered from the three examples above, in cases
of knowledge, we want one’s beliefs to be responsive to the facts. Accordingly, we
claimed that only prima facie cognitive processes can be knowledge-conducive, but
not all of them will do. The process must be a cognitive ability, meaning that the agent
will be able to be conscientious and thereby subjectively justified in employing it. And
in the absence of positive reasons (that are accessible through reflection alone) for the
reliability of the cognitive process, this last condition can be satisfied in a realistic way
only if the relevant process is a normal disposition or habit of the agent such that it
can become part of (or be integrated into) the rest of her cognitive system.
Notably then, the general idea, which all the above considerations are alluding to,
is that for a reliable process to be knowledge-conducive it must be a cognitive ability.
This idea has also appeared in the literature as the ability intuition on knowledge and
can be summarized by stating that knowledge is belief that is true in virtue of cognitive

9 An anonymous referee is worried that I should not use ‘dispositions’ and ‘habits’ as synonyms. Specifi-
cally, not all dispositions are habits; someone or something, for example, may be disposed to act in a certain
way—should the appropriate conditions obtain—even if the relevant person or thing has never behaved in
that way before. Accordingly, the worry further goes, the fact that a cognitive ability is a disposition does
not mean it will also be a habit. In response, even though it is true that in one sense of the term, ‘dispositions’
are not always going to be habits, there is another sense of the term that they are; according to this second
sense of the term, to claim that cognitive abilities are dispositions means that abilities are character traits, or
habitual behaviors that the agent tends to exhibit. A strong indication that this is how we should understand
the dispositional nature of cognitive abilities has to do with the fact that abilities can only be acquired and
sustained through practice, whereas dispositions, in the other meaning of the term, can be possessed by
an entity even if they are never actually manifested (e.g., a vase may be fragile even if it has never been
broken). As we shall see below, Greco appears to concur with this understanding of abilities as he claims
they are the stable traits of the agent’s cognitive character; a behavior can be in character only of it is
habitually manifested. See also (Greco 2010, p. 150). I am thankful to the referee for pressing this point.
10 I here say ‘the only realistic way’ because we can imagine, for instance, a case of a benevolent mentalist
who hypnotizes the agent to trust a newly acquired process, and trust it only in the appropriate conditions
(thereby allowing him to be epistemically responsible in employing it), despite the fact that the process is
not a disposition of hers.
11 For further discussion of the above intuitions on the Serendipitous Brain Lesion and Careless Math
Student cases see (Greco 1999) and (Greco 2010, pp. 149–155). For the discussion of similar thought
experiments and intuitions see (BonJour 1980), (Goldman 1986), (Lehrer 1990), and (Plantinga 1993).

123
Synthese

ability.12 Now the end result of the above considerations is that they fill in the details of
which reliable processes may plausibly count as cognitive abilities by demonstrating
that it is only normal, dispositional or habitual cognitive processes that the agent can
be subjectively justified in employing (and thereby able to gain knowledge from).
Now, building on considerations very similar to the above ones, Greco (1999)
has proposed a virtue reliabilist account of knowledge, which emphasizes that when
we assess whether some agent knows, we shouldn’t be focusing on the reliability of
isolated (cognitive) belief-forming processes, but on the reliability of the overall agent,
conceived of as a stable, interconnected system of such belief-forming processes.13 It
is this interwoven totality of cognitive abilities that give rise to one’s sense of epistemic
self (or in Greco’s terms to one’s ‘cognitive character’), and which should be the focus
of our epistemic assessment.
To make apparent the motivation for this change of epistemic focus, remember
that in order to avoid the Humean problematic we must accommodate subjective
justification in a way that does not involve knowledge or even beliefs about reliability.
At this point, Greco (1999, p. 289) suggests that a promising strategy for doing so is
to claim that “a belief p is subjectively justified for a person S (in the sense relevant
for having knowledge) if and only if S s believing p is grounded in the cognitive
dispositions that S manifests when S is thinking conscientiously” (i.e., when S is
motivated to believe what is true). In this way, the agent will employ his reliable
cognitive processes only in circumstances that have not been problematic in the past,
and he will be able to do so without even having any beliefs about their reliability.14
Greco, then, goes on to further claim that the dispositions/habits that a person manifests
when she is thinking conscientiously intertwine with each other and give rise to what
we may call one’s cognitive character (Greco 1999, p. 290). So, overall, “a belief p

12 The idea that knowledge must be grounded in cognitive abilities can be traced back to the writings of
(Sosa 1988, 1993) and Plantinga (1993). For more recent approaches to this intuition, see Greco (1999;
2004; 2007) and Pritchard (2009, 2010a, 2010b, 2012).
13 Any theory of knowledge that places in its center the ability intuition on knowledge will fall under
the general trend of Virtue Reliabilism (abilities are normally understood as virtues and vice versa). To
accentuate the features of his account, Greco calls his view Agent Reliabilism, but it is clearly a version
of Virtue Reliabilism—one that emphasizes the importance of the overall agent in the manifestation of the
relevant intellectual virtues.For alternative, robust as well as weaker formulations of Virtue Reliabilism see
(Sosa 1993, 2007) and Pritchard (2010a, 2010b, 2012), respectively.
14 The fact that people manifest highly specific, finely tuned dispositions to form their beliefs in certain
ways but not in others amounts to an implicit awareness of the reliability of those dispositions.

For example suppose that it seems visually to a person that a cat is sleeping on the couch, and on this
basis she believes that there is a sleeping cat on the couch. Suppose also that this belief manifests a
disposition that the person has, to trust this sort of experience under these sorts of conditions, when
motivated to believe the truth. Now, suppose that much less clearly, it seems visually to the person
that a mouse has run across the floor. Not being disposed to trust this kind of fleeting experience,
the person refrains from believing until further evidence comes in. The fact that the person, properly
motivated, is disposed to trust one kind of experience but not the other, constitutes sensitivity on her
part that the former is reliable. There is a clear sense in which she takes the former experience to be
adequate to her goal of believing the truth, and takes the latter experience not to be. And this is so
even if she has no beliefs about her goals, her reliability, or her experience (Greco 1999, p. 290 ).
A similar argument can be found in (Sosa 1993, pp. 60–63).

123
Synthese

has a positive epistemic status for a person S just in case S’s believing p results from
the stable and reliable dispositions that make up S s cognitive character” (ibid., pp.
287–8).
In this way, Greco can do away both with strange and fleeting processes. Strange
processes cannot be part of the agent’s cognitive character because they are not the
kind of processes that a conscientious agent would employ. Fleeting processes are
also excluded. First, because they are not dispositions or habits—so they cannot really
count as character traits. And, second (I may add), because, in the absence of reasons
to believe that the relevant process is reliable, it is only dispositions or habits that
one can become aware they are unreliable in certain circumstances and, so—without
relying on any beliefs about their reliability—use them conscientiously in the rest of
the circumstances.15
So in order to gain knowledge on the basis of a process the agent must be able
to employ a conscientious attitude towards that process. And in order for that to be
the case, the relevant process must be a cognitive ability of the agent, meaning that
it must have been integrated into the agent’s cognitive character. Now, despite the
previous points on the importance of the normality and dispositionality of the relevant
process in order for it to count as a genuine part of one’s cognitive character, Greco
attempts to further accentuate and shed some light on the integrated nature of our
cognitive characters by noting that the process of “cognitive integration is a function
of cooperation and interaction, or cooperative interaction with other aspects of the
cognitive system” (2010, p. 152). So, how exactly should we think about the required
conditions for a process to count as knowledge-conducive?
In general, every knowledge-conducive process must be a cognitive ability such
that the agent will be subjectively justified in employing it, which requires that the
process be integrated into the agent’s cognitive character by cooperatively interacting
with it. Accordingly, we may say that the only necessary and sufficient condition for
a process to count as knowledge-conducive is that it cooperatively interacts with the
rest of the agent’s cognitive character. Now, apparently, this makes the normality and
dispositionality criteria seem redundant—which strictly speaking they are—but they
may still have a role to play; normality and dispositionality of the relevant process
seem to be practical preconditions for the agent to be able to cooperatively interact
with it. The extent, however, to which each one of these criteria may need to be satisfied
will differ from case to case. An agent, for example, may be subjectively justified in
employing a process in the appropriate conditions not because it is a normal disposition
of hers but because a benevolent mentalist hypnotized her to do so. In most realistic
cases, however, normality and dispositionality will still have a significant guiding
effect. The decisive effect of cognitive integration, however—on the basis of which
the agent can be conscientious and thereby subjectively justified—will only be ensured
if the agent’s cognitive character mutually interacts with the relevant process. So, all
we need to accept that a process is knowledge-conducive is that it be integrated into

15 Apart from the example given in the previous footnote, Greco has not attempted to provide an account of
how the process of subjective justification works. I assume, however, that he wouldn’t reject this falsifica-
tionist approach, as I cannot see how else subjective justification could be accommodated in an externalist
way.

123
Synthese

the agent’s cognitive character by cooperatively interacting with it. On the basis of this
mutual interaction with the rest of his cognitive system the agent will be able to employ
the relevant process conscientiously by merely lacking any beliefs that it is unreliable
(at least not in the circumstances in which he employs it), or if the employment is
involuntary, conscientiously accept its deliverances when he lacks any beliefs that the
conditions that gave rise to them were unreliable.
Moreover, as we saw previously in the discussion of the three examples, this process
of cognitive integration gives rise to a coherentist effect both on the level of processes
(how the beliefs are generated) and on the level of content (how the beliefs themselves
combine). Also, it ensures that at least directly related belief-generating mechanisms
and their resulting beliefs will be consistent.
Overall then, the epistemic importance of cognitive integration is that it allows
epistemic agents to satisfy the condition of subjective justification in a minimalist way,
which is nevertheless sufficient for acquiring knowledge even if one cannot—not even
in principle—offer any explicit reasons in favor of one’s beliefs. Stated explicitly, this
minimalist condition of subjective justification is that conscientious epistemic agents
ought to accept the deliverances of, and employ their cognitive abilities, only when
they lack any doubts that they are unreliable, given the conditions they employ them
in.16 In other words, cognitive integration allows epistemic agents to be conscientious
in the sense that in cases where there is something wrong with the way they form their
beliefs they will be able to spot this and respond appropriately. Otherwise, they can go
on formulating their beliefs, without worrying whether they can actually offer reasons
for every single one of them or for their reasons for holding them.17

16 As an anonymous referee has pointed out I should make clear that this condition should be restricted
to reliable processes. We should not allow, for example, to an agent who forms his beliefs on the basis of
astronomical considerations or wishful thinking to count as subjectively justified merely by lacking any
doubts about the unreliability of his belief-forming processes. Given, however, that the epistemic agent
must be conscientious (i.e., motivated to believe what is true) this qualification is actually redundant. If
the agent is motivated to believe what is true, he will not employ astronomical considerations or wishful
thinking because he will have noticed that such process were notably unreliable in the past. For the same
reason, they won’t even be parts of his (conscientious) cognitive character.
17 Further to footnotes 14 and 15, in providing this sort of account of subjective justification, I have
relied for the most part on phenomenological intuitions about how we seem to go about our beliefs in
everyday life. Nevertheless, such phenomenological intuitions seem to already entertain a certain degree
of scientific support. Specifically, within cognitive psychology, there have been several studies indicating
that subjects engage in analytic reasoning only when they experience the metacognitive effect of the lack of
‘fluency’:

“Fluency is not a cognitive operation in and of itself but, rather, a feeling of ease associated with a
cognitive operation, it can be generated by nearly any form of thinking. If a percept is blurry, we
are aware that it was hard to see. If a word is phonemically irregular, we recognize the challenge
in processing it. We know whether we had to struggle to bring a memory to mind and whether we
had a hard or easy time solving a riddle. Because the metacognitive experience of fluency can be
generated by so many cognitive processes and is nearly effortless to access, it can serve as a cue
toward judgments in virtually any situation”. (Oppenheimer 2008).
For an overview on the metacognitive feeling of fluency see (Oppenheimer 2008), (Alter and Oppenheimer
2007) and (Unkelbach and Greifeneder 2013).

123
Synthese

2.3 Cognitive integration in epistemology and philosophy of mind

In the introduction, I mentioned that the topic of cognitive integration has been explored
in epistemology also in reference to the possibility of cognitive extension. The general
idea is that the knowledge-conducive cognitive abilities and their relation to one’s
cognitive character as discussed within virtue epistemology is particularly apt for an
interpretation along the lines suggested by the hypothesis of extended cognition. To
make this clear it should be helpful to repeat, for the last time, what are the important
features of a knowledge-conducive belief-forming process.
In general, the process must be a cognitive ability. In order for that to be the case
the process must be a cognitive process. This will guarantee the correct direction
of fit between the belief and the fact. Also, we want the belief-forming process to be
objectively reliable, where a reliable process is one that tends to produce true rather than
false beliefs. Recall, however, that according to reliabilism, the agent does not need his
evidence to be necessarily reliable, such that he can be internally justified in holding
his belief; if forming a belief on a certain kind of evidence constitutes a reliable belief-
forming process, it does not matter that one’s evidence is only contingently reliable; the
agent, on his part, does not need to know or even have any beliefs about the reliability
of his way of forming beliefs. Instead, the agent can be subjectively justified simply
by forming his belief on the basis of a process that is integrated into his cognitive
character, which he employs when he is thinking conscientiously. Now, in order for a
process to be a candidate for inclusion to the agent’s conscientious cognitive character,
we noted that it will probably have to be neither strange nor fleeting. In most realistic
scenarios the process (1) will have to be normal so that the agent won’t reject it when
conscientious and (2) will have to be a disposition or a habit of the agent, because
(barring scenarios such as the mentalist case) it is only dispositions or habits that
one can become aware they are unreliable in certain circumstances, and so—without
relying on any beliefs about their reliability—be able to employ them conscientiously
in the rest of the circumstances. As we further noted, however, even though normality
and dispositionality will, in most cases, be practical preconditions, they are neither
necessary nor sufficient for a process to count as integrated into the agent’s cognitive
character. Instead, the only requirement is that the process be integrated into the agent’s
cognitive character, by engaging in cooperative interaction with the rest of the agent’s
cognitive system. Accordingly, no matter what the practical preconditions for this
interactive process to be achieved are, once it is in place, it will guarantee both that
the relevant belief-forming process is a cognitive process, and that it is indeed part of
the agent’s cognitive character such that he can be conscientious in employing it. So
putting all the above points together: a belief-forming process counts as a cognitive
ability and thereby as knowledge-conducive if and only if it is a reliable belief-forming
process that is integrated into the agent’s cognitive character, on the basis of a process
of cooperative interaction with it.18

18 Many externalist epistemologists would reject the above biconditional on the grounds that in order for a
process to be knowledge-conducive it should also be safe (where a safe process is one that could not have
easily being wrong). Consider for example Anti-Luck Virtue Epistemology: S knows that p if and only if S’s
safe belief that p is the product of her relevant cognitive abilities (such that her safe cognitive success is to

123
Synthese

Now to see why we might get the impression of a close fit between virtue epis-
temology and the hypothesis of extended cognition, here are three common-sense
functionalist criteria, which Clark (2010) suggests must be satisfied by non-biological
candidates in order to be included into an individual’s cognitive system:

(1) “That the process be reliably available and typically invoked”.

That is, the agent should habitually and easily invoke the external resource. In other
words, its employment must be a disposition/habit of the agent’s overall cognitive
mechanism.

(2) “That any information thus retrieved be more-or-less automatically endorsed.

It should not usually be subject to critical scrutiny. […] It should be deemed about
as trustworthy as something retrieved clearly from biological memory”.

That is, the information in the resource must be regarded as normal and reliable
and not be necessarily reliable. It suffices that its employment result into an equally
trustworthy belief-forming process as the one of forming beliefs on the basis of one’s
own biological memory.19
At this point, however, one might object that being reliable is not the same as
being trustworthy (i.e., being regarded as reliable). But, in response, notice first that
Clark identifies the notion of trustworthiness of a process with the idea of being
“more-or-less automatically endorsed” or in other words “not usually subject to critical
scrutiny”. That is, the target process must not have been (for the most part) problematic
in the past. Moreover, the processes under consideration are also supposed to be
cognitive dispositions or habits of the agent that he has repeatedly employed in the
past, and so had they been problematic the agent would have noticed that and responded
appropriately. Accordingly, a trustworthy belief-forming process in Clark’s account,
will be one that tends to produce true rather than false beliefs, which is to say that it
will be objectively reliable in the virtue reliabilist’s sense. What the agent will deem
reliable will be that which is objectively reliable, i.e., that which has not been (for the
most part) problematic in the past.
Furthermore, notice this negative way of deeming processes reliable with which
Clark concurs (i.e., that a trustworthy process is one that is not usually subject to critical
scrutiny such that it is more-or-less automatically endorsed). What this means is that
the agent does not need to have any beliefs about why or whether his belief-forming
process is trustworthy; it suffices that it has not repeatedly caught his negative attention
in the past. This is in good agreement with the proposed minimalist understanding of
subjective justification according to which one does not need to rely on any beliefs but

Footnote 18 continued
a significant degree creditable to her cognitive agency) (Pritchard 2012, p. 20). Again, in Pritchard (2010a,
p. 76) we can read: “ knowledge is safe belief that arises out of the reliable cognitive traits that make up one’s
cognitive character, such that one’s cognitive success is to a significant degree creditable to one’s cognitive
character”. For a defense of the claim that the safety condition is not necessary for virtue reliabilism to
account for knowledge see (Palermos forthcoming).
19 That is, the process does not need to be, due to underlying logical or quasi-logical relations, 100 %
reliable. Notice that memory is supposed to be reliable even though one may misremember.

123
Synthese

simply on one’s motivation to believe the truth. For example, one will trust one’s vision
in appropriate circumstances, just because vision has not been notably problematic
in the past (in those circumstances). By being motivated to believe the truth one will
thereby employ the belief-forming process that has not in the past (notably) failed to
be conducive towards that end, and crucially, one will do so without even thinking
about it.
(3) “That information contained in the resource should be easily accessible as
and when required”.
That is, the agent must be able to employ it as if it was part of his organismic cognitive
mechanism. In other words, the resource must be integrated into the agent’s overall
cognitive mechanism.
So we see that the same features of a process that epistemologists deem important
in order for a process to be knowledge-conducive are required by a common-sense
functionalist understanding of cognition in order for a process to count as part of one’s
mind. This is a promising observation.
Notice, however, that even if there were no such close fit between these broad
features, we would still be able to show that the two theories are essentially con-
nected. The reason, as we noted before, is that some of the above features (e.g.,
normality and dispositionality of the process) may be conducive towards the process
being knowledge-conducive, but they are neither necessary nor sufficient for that end.
Instead, the only requirement is that the process be integrated into the agent’s cognitive
character.
Now in relation to this, philosophy of cognitive science has recently started shifting
its focus away from the above common-sense functionalist criteria for including an
external resource into one’s cognitive system. Instead, Froese et al. (2013), Chemero
(2009) and Palermos (2014) have suggested that the only requirement for an external
element to count as a constitutive part of the agent’s cognitive system is that it be
non-linearly related to the rest of the agent’s cognitive system. The motivation for this
is that, according to dynamical systems theory, these non-linear relations give rise to
an overall non-decomposable system that consists of all the contributing parts. And
two reasons for postulating the overall system are that these non-linear interactions
(1) give rise to new systemic properties that belong only to the overall system and
to none of the contributing systems alone (therefore we have to postulate the overall
extended system) and (2) prevent us from decomposing the two systems in terms of
distinct inputs and outputs from the one subsystem to the other (therefore we cannot but
postulate the overall system).20 What is even more interesting to our present purposes,
however, is that just as Greco holds that cognitive integration is a matter of interaction
and cooperation between cognitive processes, so those non-linear relations that allow
us to talk about integration within philosophy of mind and cognitive science emerge
only on the basis of cooperative feedback loops between the contributing elements of
the overall system.

20 For a detailed explanation of why the existence of non-linear relations that arise out of the mutual
interactions between agents and their artifacts ensures the existence of extended cognitive systems see
(Palermos 2014).

123
Synthese

Therefore both in epistemology and philosophy of mind and cognitive science the
same criterion (cooperative interaction with the rest of the agent’s cognitive system)
is required for a process to be integrated into an agent’s cognitive system and thereby
count as knowledge-conducive. This, however, should not really come as a surprise.
Given that virtue reliabilism holds that knowledge must be the product of cognitive
ability (however that ability may be realized) and that the hypothesis of extended
cognition sets out to reveal which processes can count as cognitive abilities (wherever
they may be located), this close fit between the two theories seems to be as it should
be.
The conclusion that follows, then, is that there is no principled theoretical bar dis-
allowing extended belief-forming processes from counting as knowledge-conducive
cognitive abilities. Given that virtue reliabilism makes no specifications as to whether
knowledge-conducive cognitive abilities should be located within the agent’s head,
then, provided that the condition of cognitive integration is met, the epistemic agent
may extend his knowledge-conducive cognitive character beyond his organismic cog-
nitive abilities by incorporating epistemic artifacts to it.

3 Is this necessary for epistemology and what are the ramifications?

Obviously, the possibility of knowledge-conducive cognitive characters that may nev-

ertheless be extended beyond our organismic cognitive capacities can generate inter-
esting ramifications both for traditional and social epistemology, which may allow us
to think about knowledge in new ways. Focusing, however, on the integrated nature of
our extendable cognitive characters may not only be important for moving forward,
but also necessary for accounting for knowledge as we already think about it.
According to Greco (1999, p. 287), in addition to one’s organismic cognitive
abilities of the brain/central nervous system, a person’s cognitive character may also
consist of “acquired skills of perception and acquired methods of inquiry including
those involving highly specialized training or even advanced technology”. The reason
for this move is that we need to account for advanced cases of knowledge where
one’s believing the truth is the product of the operation of epistemic artifacts such as
telescopes, microscopes, tactile visual substitution systems and so on.21 The problem,
however, is that in the traditional conception, cognition takes place strictly within the
agent’s head and so artifacts cannot be parts of one’s cognitive character.
One way to sidestep this problem for virtue reliabilism, could be to claim that, in
such cases, it is merely the agent’s training and skill of using the artifact, as mirrored
in the agent’s neural/bodily architecture, that is the most salient factor in the causal
explanation of the agent’s cognitive success (i.e., believing the truth). Notice, however,
that when an agent employs an epistemic tool, his true belief arises as the product of the
interaction between his internal processes and the artifact. According to dynamical
systems theory, then, the cognitive process that allows the agent to detect the truth
is not merely ‘aided’ or ‘assisted’ by the artifact but is, instead, constituted by it as

21 See Bach-y-Rita and Kercel (2003) for a recent review on tactile visual substitution systems.

123
Synthese

it arises out of the ongoing mutual interaction between the agent and the artifact.22
Therefore, in a causal explanation of how the agent acquired his true belief, it will be
impossible to disentangle the agent’s training and skill of using the artifact from his
actual engagement with it.23
But even if such decomposition were possible, notice in addition that the part of
the process that allows the agent to detect the truth, or in other words to be sensitive
to the facts, is the external component. To illustrate this, consider, on one hand, an
untrained agent in possession of a properly working artifact. In that case, it is obvious
that even though the agent will initially be unable to form any (true or false) beliefs,
eventually—provided that he gains sufficient experience such that he can interact with
it—not only will he form beliefs, but he will also reliably enjoy cognitive success. On
the other hand, think about a well-trained agent, but in possession of a faulty artifact.
In this case, despite the agent’s excellent internal skills, it is evident that he would be
unable to reach any (non-lucky) true beliefs, no matter how much he tried. It therefore
seems that in such cases the most (and maybe the only) significant factor that explains
the truth-status of the agent’s belief is the epistemic artifact. In other words, since the
agent’s belief is true in virtue of the artifact, the virtue reliabilist must account for it
being part of the agent’s cognitive system. Given, however, that cognition is normally
supposed to take place within the agent’s head, virtue reliabilists can only account for
such cases by wedding their view to the hypothesis of extended cognition. Accordingly,
combining the extended cognition hypothesis with virtue reliabilism on the basis of
their close fit does not seem to be just an available option for epistemologists, but also
necessary for dealing with advanced cases of knowledge where the latter is the product
of the employment of epistemic artifacts that the agent mutually interacts with.
Apart from the necessity of introducing the extended cognition hypothesis within
epistemology, however, this move can also generate interesting ramifications, espe-
cially for social epistemology. For instance, it can lead to the further claim that there
could be cognitive characters that do not just extend beyond an agent’s organismic
capacities, but which are instead distributed amongst several agents along with their
epistemic artifacts. The hypothesis of distributed cognition, which has been developed
in parallel to the hypothesis of extended cognition (Hutchins 1995; Theiner et al. 2010;
Sutton et al. 2008; Wilson 2005; Heylighen,Heath,Van Overwalle 2007), differs to the
latter position only in that this time the cognitive system extends to include epistemic
artifacts as well as other agents. And interestingly, most proponents of the view (Sutton
et al. 2008; Theiner et al. 2010; Wegner et al. 1985; Tollefsen and Dale 2011) point
out again that it is the existence of non-linear cooperative interactions between the
contributing members and their artifacts that is the criterion by which we can judge

22 It should be here noted that not every case of the employment of an artifact is a case of cognitive
extension, but only when the agent mutually interacts with it. For an objective criterion of constitution and
on what may count as a genuine case of cognitive extension, see (Palermos 2014).
23 Remember that according to virtue reliabilism and the underlying ability intuition on knowledge, knowl-
edge is belief that is true in virtue of cognitive ability, where, according to Greco, “in virtue of” must be
understood in causal explanatory terms. Even though several proponents of virtue reliabilism agree on this
general causal-explanatory understanding of the view, there is disagreement on whether the relevant cog-
nitive ability should be the “most salient” (Greco 2010) or merely a “significant” (Pritchard 2010b) factor
in the causal explanation of how the agent acquired his true belief.

123
Synthese

whether we have an integrated distributed cognitive system. Accordingly, there could

be knowledge-conducive cognitive characters, which may nevertheless be distributed.
This is an interesting possibility, because it can allow us to combine an individualis-
tic approach to knowledge, such as virtue reliabilism, with the hypothesis of distributed
cognition in order to account for epistemic group agents: Groups of individuals who
exist and gain knowledge in virtue of a shared, common cognitive character that mainly
consists of a distributed cognitive ability—a collective cognitive ability that emerges
out of the members’ mutual (socio-epistemic) interactions and which is not reducible
to the cognitive abilities possessed by the individual members, thereby allowing us to
speak of a group agent in itself. This is important, because by recognizing a group of
people as a self-standing agent in itself, we can then use an individualistic approach
to knowledge to account for knowledge that is collectively produced and which is,
thereby, distinctively social. In other words we can make sense of the claim that p is
known by S (the group agent), even though it is not known by any individual alone.24
Remarkably, such group agents have already started being studied within cognitive
science. Consider for example transactive memory systems (TMSs)—i.e., groups of
two or more individuals who collaboratively encode, store and retrieve information.
The reason why TMSs are good candidates for distributed cognitive systems—and
thereby for epistemic group agents—is that, as Sutton et al. observe (2008), such
systems are likely to involve skillful interactive simultaneous coordination of people
who can thereby count as a single integrated cognitive system. Therefore, we can use
TMSs in order to “conceptualize how people in close relationships may depend on
each other for acquiring, remembering, and generating knowledge” (Wegner et al.
1985, p. 253):
Ordinarily psychologists think of memory as an individual’s store of knowledge,
along with the processes whereby that knowledge is constructed, organized, and
accessed. […] With transactive memory we are concerned with how knowledge
enters the dyad, is organized within it, and is made available for subsequent use
by it (ibid., 256).
Apart from incorporating already existing research from cognitive science, however,
the combination of virtue reliabilism with the hypotheses of extended and distributed
cognition can generate new avenues for research, some of which have for a long time
been inaccessible. An interesting example is the intersection between epistemology
and the field of history and philosophy of science. These two intimately related fields
have so far been at odds—an awkward situation owing to the fact that the former disci-
pline has traditionally being individualistic whereas the latter has for the most part been
socially oriented (hardly anyone could deny the social nature of the scientific process,
especially after the publication of Kuhn’s The Structure of the Scientific Revolutions,
in 1962). The present approach, however, could now provide a useful link between
the two fields. Science is primarily performed by individual scientists employing their
hardware and software epistemic artifacts or by research teams operating within sci-
entific labs that are uniquely tailored to fit their purposes. Accordingly, the concepts of

24 For a more detailed explanation of how virtue reliabilism may be applied to epistemic group agents see
(Palermos and Pritchard 2013).

123
Synthese

extended cognitive characters and epistemic group agents could become very handy
for a mainstream epistemological analysis of the scientific progress. As Giere and
Moffat (2003, p. 308) note in their discussion of the scientific revolution of the 16th
century,
“No ‘new man’ suddenly emerged sometime in the sixteenth century. . .. The
idea that a more rational mind . . . emerged from darkness and chaos is too
complicated a hypothesis” [Latour 1986, 1]. We agree completely. Appeals to
cognitive architecture and capacities now studied in cognitive sciences are meant
to explain how humans with normal human cognitive capacities manage to do
modern science. One way, we suggest, is by constructing distributed cognitive
systems that can be operated by humans possessing only the limited cognitive
capacities they in fact possess.

4 Conclusion

The topic of cognitive integration is a far-reaching multidisciplinary theme that touches

upon a wide range of disparate questions with the potential of bringing several research
areas together. Within epistemology, the phenomenon of cognitive integration can
reveal how knowers may be subjectively justified despite the absence of positive rea-
sons in favor of their true beliefs, pointing towards a minimalist yet informative epis-
temic norm: provided that one’s belief-forming process has been integrated into one’s
cognitive system/character and that one is motivated to believe what is true, one can
be justified in holding the resulting belief merely by lacking any doubts that his way of
forming his belief, or that the belief itself, is inappropriate. Within philosophy of mind,
cognitive integration can explain how it is possible to extend our cognitive capacities
beyond our organisms to the artifacts we employ or other individuals we may interact
with. And combining philosophy of mind with epistemology by using the process of
cognitive integration as the connecting point provides the necessary means to account
for advanced cases of knowledge where the known belief is true in virtue of the oper-
ation of epistemic artifacts or even the activity of collaborative groups. Quite likely,
however, such an account of knowledge won’t be valuable just within epistemology,
philosophy of mind and their intersection. Focusing on the essentially technological
and collaborative nature of the scientific process, such an approach to knowledge could
finally provide a strong link between the related but so far persistently isolated fields
of philosophy of mind, philosophy of science, and epistemology.

Acknowledgments I am very thankful to John Greco and an anonymous referee for detailed comments
on pervious versions of this paper.

References

Alter, L., & Oppenheimer, D. M. (2007). Overcoming intuition: Metacognitive difficulty activates analytic
reasoning. Journal of Experimental Psychology, 136(4), 569–576.
Bach-y-Rita, P., & Kercel, S. W. (2003). Sensory substitution and the human-machine interface. Trends in
Cognitive Science 7, 541–546.

123
Synthese

BonJour, L. (1980). Externalist theories of empirical knowledge. Midwest Studies in Philosophy, 5, 53–73.
BonJour, L. (1985). The structure of empirical knowledge. Cambridge, MA: Harvard University Press.
BonJour, L. (2002). Internalism and externalism. In P. Moser (Ed.), Oxford handbook of epistemology (pp.
234–264). Oxford: Oxford University Press.
Brewer, F. W., & Lambert, B. L. (2001). The theory ladeness of observation and the theory: Ladeness of
the rest of the scientific process. Philosophy of Science. In Proceedings of the 2000 Biennial Meeting of
the Philosophy of Science Association. Part I: Contributed Papers (Vol 68(3), pp. S176–S186).
Clark, A., & Chalmers, D. (1998). The extended mind. Ananlysis, 58(1), 7–19.
Clark, A., & Chalmers, D. (2008). Supersizing the mind. New York: Oxford University Press.
Clark, A., & Chalmers, D. (2010). Memento’s revenge: The extended mind, extended. In R. Menary (Ed.),
The extended mind. Cambridge, MA: MIT press.
Chemero, A. (2009). Radical embodied cognitive science. Cambridge, MA: MIT press.
Chisholm, R. M. (1977). Theory of knowledge (2nd ed.). Englewood Cliffs, NJ: Prentice-Hall.
Churchland, P. M. (1979). Scientific realism and the plasticity of the mind. Cambridge: Cambridge University
Press.
Churchland, P. M. (1988). Perceptual plasticity and theoretical neutrality: A reply to Jerry Fodor. Philosophy
of Science, 55, 167–187.
Churchland, P. M. (1989). A neurocomputational perspective. The nature of mind and the structure of
science. Cambridge, MA: The MIT Press.
Estany, A. (2001). The thesis of theory-Laden observation in the light of cognitive psychology. Philosophy
of Science, 68(2), 203–217.
Feyerabend, P. K. (1975). Against method: Outline of an anarchistic theory of knowledge. London: New
Left Books.
Fodor, J. A. (1983). The modularity of mind. Cambridge, MA: MIT Press.
Fodor, J. A. (1984). Observation reconsidered. Philosophy of Science, 51, 23–43.
Fodor, J. A. (1988). A reply to Churchland’s: Perceptual plasticity and theoretical neutrality. Philosophy of
Science, 55, 188–198.
Froese, T., Gershenson, C., & Rosenblueth, D. A. (2013). The dynamically extended mind. http://arxiv.org/
abs/1305.1958.
Fuller, S. (2012). Social epistemology: A quarter-century itinerary. Social Epistemology: A Journal of
Knowledge, Culture and Policy, 26, 267–283.
Giere, R., & Moffat, B. (2003). Distributed cognition: Where the cognitive and the social merge. Social
Studies of Science, 33(2), 1–10.
Goldman, A. (1986). Epistemology and cognition. Cambridge, MA: Harvard University Press.
Greco, J. (1999). Agent reliabilism. In J. Tomberlin (Ed.), Philosophical perspectives 13: Epistemology
(pp. 273–296). Atascadero, CA: Ridgeview Press.
Greco, J. (2004). Knowledge as credit for true belief. In M. DePaul & L. Zagzebski (Eds.), Intellectual
virtue: Perspectives from ethics and epistemology. Oxford: Oxford University Press.
Greco, J. (2007). The nature of ability and the purpose of knowledge. Philosophical Issues, 17, 57–69.
Greco, J. (2010). Achieving knowledge: A virtue-theoretic account of epistemic normativity. Cambridge:
Cambridge University Press.
Haddock, A., Millar, A., & Pritchard, D. H. (2010). Knowledge and understanding. The nature and value
of knowledge: Three investigations. Oxford: Oxford University Press.
Hanson, N. R. (1961). Patterns of discovery. Cambridge: Cambridge University Press.
Hanson, N. R. (1969). Perception and discovery: An introduction to scientific inquiry. San Francisco, CA:
Freeman Cooper.
Heylighen, F., Heath, M., & Van Overwalle, F. (2007). The emergence of distributed cognition: A conceptual
framework. In Proceedings of Collective Intentionality IV, Vol 4. University of Siena.
Hutchins, E. (1995). Cognition in the wild. Cambridge, MA: MIT Press.
Kuhn, T. S. (1962). The structure of scientific revolutions. Chicago, IL: The University of Chicago Press.
Lehrer, K. (1990). Theory of Knowledge. Boulder, CO: Westview.
Oppenheimer, D. (2008). The secret life of fluency. Trends in Cognitive Science, 12(6), 237–241.
Palermos, S. O. (forthcoming). Could reliability naturally imply safety? European Journal of Philosophy.
doi:10.1111/ejop.12046.
Palermos, S. O. (2011) Belief-forming processes, extended. Review of Philosophy and Psychology, 2, 741–
765.
Palermos, S. O. (2014). Loops, constitution and cognitive extension, Cognitive Systems Research, 27, 25–41.

123
Synthese

Palermos, S. O., & Pritchard, D. (2013). Extended knowledge and social epistemology. Social Epistemology
Review and Reply Collective, 2(8), 105–120.
Plantinga, A. (1993a). Warrant and proper function. New York: Oxford University Press.
Plantinga, A. (1993b). Warrant: The current debate. Oxford: Oxford University Press.
Pappas, G. (2005). Internalist versus externalist conceptions of epistemic justification. In E. Zalta (Ed.)
Stanford encyclopaedia of philosophy. http://plato.stanford.edu/entries/justep-intext/.
Poston, T. (2008). Internalism and externalism in epistemology. In B. Dowden & J. Fieser (Eds.) Internet
encyclopaedia of philosophy. http://www.iep.utm.edu/int-ext/.
Pritchard, D. ( forthcoming). Anti-luck virtue epistemology. Journal of Philosophy.
Pritchard, D. (2009). Knowledge. London: Palgrave Macmillan.
Pritchard, D. (2010). Cognitive ability and the extended cognition thesis. Synthese, 175(1), 133–151.
Pritchard, D. (2012). Anti-luck virtue epistemology. Journal of Philosophy, 109, 247-79.
Pryor, J. (2001). Highlights of recent epistemology. British Journal for the Philosophy of Science, 52,
95–124.
Sosa, E. (1988). Beyond skepticism, to the best of our knowledge. Mind, New Series, 97(386), 153–188.
Sosa, E. (1993). Proper functionalism and virtue epistemology. Nous, 27(1), 51–65.
Sosa, E. (2007). A virtue epistemology: Apt belief and reflective knowledge. Oxford: Clarendon Press.
Steup, M. (1999). A defense of internalism. In L. Pojman (Ed.), The theory of knowledge: Classical and
contemporary readings (pp. 310–321). Belmont, CA: Wadsworth.
Sutton, J., Barnier, A., Harris, C., & Wilson, R. (2008). A conceptual and empirical framework for the social
distribution of cognition: The case of memory. Cognitive Systems Research, 9(1), 33–51.
Theiner, G., Allen, C., & Goldstone, R. (2010). Recognizing group cognition. Cognitive Systems Research,
11(4), 378–395.
Tollefsen, D., & Dale, R. (2011). Naturalizing joint action: A process-based approach. Philosophical Psy-
chology, 25(3), 385–407.
Unkelbach, C., & Greifeneder, R. (2013). A general model of fluency effects in judgment and decision
making. In C. Unkelbach & R. Greifeneder (Eds.), The experience of thinking (pp. 11–32). New York:
Psychology Press.
Wegner, M., Giuliano, T., & Hertel, P. (1985). Cognitive interdependence in close relationships. In W. J.
Ickes (Ed.), Compatible and incompatible relationships (pp. 253–276). New York: Springer.
Wilson, R. A. (2005). Collective memory, group minds, and the extended mind thesis. Cognitive Processing,
6(4), 227–236.

123