Philosophical Psychology, 2014

Vol. 27, No. 1, 65–81, http://dx.doi.org/10.1080/09515089.2013.828376

Emotions beyond brain and body

Achim Stephan, Sven Walter, and Wendy Wilutzky

The emerging consensus in the philosophy of cognition is that cognition is situated, i.e.,
dependent upon or co-constituted by the body, the environment, and/or the embodied
interaction with it. But what about emotions? If the brain alone cannot do much thinking,
can the brain alone do some emoting? If not, what else is needed? Do (some) emotions
(sometimes) cross an individual’s boundary? If so, what kinds of supra-individual systems
can be bearers of affective states, and why? And does that make emotions “embedded” or
“extended” in the sense cognition is said to be embedded and extended? Section 2 shows
why it is important to understand in which sense body, environment, and our embodied
interaction with the world contribute to our affective life. Section 3 introduces some key
concepts of the debate about situated cognition. Section 4 draws attention to an important
disanalogy between cognition and emotion with regard to the role of the body. Section 5
shows under which conditions a contribution by the environment results in non-trivial
cases of “embedded” emotions. Section 6 is concerned with affective phenomena that seem
to cross the organismic boundaries of an individual, in particular with the idea that
emotions are “extended” or “distributed.”
Keywords: Cognitivism; Distributed Cognition; Emotions; Enactivism; Extended
Cognition

1. Introduction
Cognitivists regard the human mind as an information-processing input-output device
whose neuronally implemented, syntactically driven transformations of represen­
tational structures give rise to cognitive processing. For them, cognition is the amodal
intracranial “filling” mediating offline between input from and output to the
extracranial parts of the body and the extrabodily environment. Situated approaches, in
contrast, pivot on the fact that cognition is primarily based on reciprocal real-time

Achim Stephan is Professor of Philosophy of Cognition at the Institute of Cognitive Science, University of
Osnabrück.
Sven Walter is Professor of Philosophy of Mind at the Institute of Cognitive Science, University of Osnabrück.
Wendy Wilutzky is a Ph.D. Student at the Institute of Cognitive Science, University of Osnabrück.
Correspondence to: Sven Walter, University of Osnabrück, Institute of Cognitive Science, Albrechtstraße 28,
Osnabrück 49069, Germany. Email: [email protected]

q 2013 Taylor & Francis

66 A. Stephan et al.
interactions of embodied agents with their environments.1 Appreciating the
contribution of body, environment, and the interaction with it is said to yield
important novel insights into the developmental and material basis of our cognitive life.
The emerging consensus is that just as we cannot do much carpentry with our bare
hands, there is not much thinking we can do with our brain alone (Dennett, 2000, p. 17).
But what about emotions? If the brain alone cannot do much thinking, can the brain
alone do some emoting, as some have claimed? If not, what else is needed? Do (some)
emotions (sometimes) cross an individual’s boundary? If so, what kinds of supra­
individual systems can be said to be bearers of affective states, and why? And does that
make emotions “embedded” or “extended” in the sense cognition is said to be embedded
and extended? The role of the body and the (mainly cultural and social) environment
has been a principal topic in emotion research for decades, but not under the heading of
“situated cognition.” We believe that applying the concepts from the situated cognition
debate to emotions can open up new avenues of research in the philosophy of emotions.
But we will also see that there is still a lot to be done before we get a firm grip on the idea
of “situated affectivity” and its various variants. This paper attempts to clear some
conceptual ground for future endeavors into this exciting new area of research.
Section 2 shows why it is important to understand in which sense body,
environment, and our embodied interaction with the natural and social world
contribute to our affective life. Section 3 sets the scene by introducing some key
concepts of the debate about situated cognition. Section 4 draws attention to an
important disanalogy between cognition and emotion with regard to the role of the
body. Section 5 shows under which conditions a contribution by the environment
results in non-trivial cases of “embedded” emotions. Finally, Section 6 is concerned
with affective phenomena that seem to cross the organismic boundaries of an
individual, in particular with the idea that emotions are “extended” or “distributed.”
We conclude that acknowledging the essentially situated character of some affective
phenomena undoubtedly enriches debates in the philosophy of emotions that have
hitherto focused too narrowly on certain individualistic phenomena. Some emotions
are dynamic engagements with the world rather than inner snapshot responses to
external triggers, and some (“joint emotions” and “affective atmospheres”) are
brought about by groups of interacting individuals.

2. Why Situated Affectivity?

Why even raise the issue of situated affectivity? Since the debate about situated
cognition is already controversial and muddled enough, applying the conceptual
schemes of that debate, under-developed as they still are, to an apparently even more
intricate field such as emotions may seem ambitious. But there are good reasons for
trying to do so.
Situated approaches to cognition are motivated by the insight that we are not
isolated quasi-Cartesian minds housed in a container-like body they intelligently
navigate through their environment by repeated sense-think-act cycles. This insight
 Philosophical Psychology 67
applies mutatis mutandis to emotions. It is not just that we aren’t isolated thinkers; we
aren’t isolated “emoters” either. Our affective life is not “sandwiched” between
perception as the input and action as the output of repeated sense-appraise-feel-act
cycles,2 and not detached from our embodied interaction with our environment.
If anything, the case for an intimate coupling of brain, body, and environment should
have been even more obvious for emotions from the very start. If thinking about
cognition had not made us wonder about situatedness, emotions should have.
Moreover, a focus on cognition alone is not going to yield a complete account of our
conditio humana, given that we were never pure animalia rationalia to begin with, but
always also animalia emotionalia and animalia motivata that do not only think, but
also feel, evaluate, care, want, and strive. The division into cognition, emotion, and
motivation as the three basic and irreducible facultates mentales that nineteenth
century faculty psychology inherited from philosophy and that still influences some
parts of cognitive science is no longer tenable. For the reasons given below, cognition
and emotion (and also motivation; see, e.g., Dai & Sternberg, 2004) are so intimately
intertwined that we should try to offer an integrated account. And if cognition is
indeed situated, there is a straightforward reason why such an integrated account is
going to be a situated one. Since most accounts of emotions acknowledge that
cognition and emotion are related and therefore in some way or other appeal to
cognitive processes, the discovery that the latter almost without exception involve, in
some yet to be specified sense, extracranial processes, is obviously going to have an
impact on our conception of emotions.
This is most obvious for cognitivist accounts of emotions that were intended as
alternatives to pure “feeling theories” à la James (1884). Whereas the latter conceived
of emotions as mere feelings of bodily changes which, as such, are not directed at
anything in the world, cognitivists regard them as worldly directed cognitive states or
processes with a specific intentional content (e.g., Nussbaum, 2001; Solomon, 1976).
Crudely put, our fear of a spider just is our belief that the spider is dangerous together
with our desire that it go away, and our anger at someone just is our judgment that we
have been wronged by that person, while the affective aspects (the unpleasant feelings
and forms of bodily arousal characteristic of fear and anger) are denigrated to mere
epiphenomena. Obviously, if the cognitive processes to which emotions are said to be
reducible are situated instead of purely intracranial transformations of represen­
tational structures, the cognitivists’ view of emotions is going to be affected as well.
Likewise, a situated view of cognition would also impact accounts that take
cognitive processes to be one of several constituents of emotions, for example
psychological appraisal theories like Scherer’s (2005) “component process model” that
treats emotions as complexes of interacting bodily, experiential, and cognitive
components, where the cognitive component is an evaluation, or appraisal, that
represents the world in the light of the specific concerns of the subject. Again, if the
cognitive appraisal-component is situated, the appraisal theorists’ view of emotions is
going to be affected as well.3
The same finally holds for those who reject Scherer’s account and similar “add-on”­
theories, as Goldie (2000) calls them, that analyze emotions in terms of discrete and in
68 A. Stephan et al.
principle separable components (where the cognitive component accounts for their
intentionality and the experiential component for their felt affectivity) in favor of a
fundamental, sui generis kind of “affective intentionality,” in which emotional feelings
are inextricably intertwined with the world-directed aspect of emotions (e.g., Slaby &
Stephan, 2008). Insofar as they acknowledge a cognitive aspect, such accounts of
emotions as cognitive-experiential “hybrids” or “felt evaluations” will also be affected,
if cognition is situated. The interplay of cognitive and bodily aspects in emotional
processes will become more comprehensible and, moreover, considering interactions
with the environment will enrich these theories.
Despite the fact that there are thus good reasons for raising the issue of situated
affectivity, it is still entirely unclear what it would mean for emotions to be situated in
the various ways in which cognition can be situated. Since a lot depends on what is
meant by calling cognition “situated,” we briefly address this issue in section 3 before
returning to affectivity.

3. Situated Cognition
“Situated cognition” is currently little more than a placeholder for a “loose-knit family
of approaches” (Wilson & Clark, 2009, p. 55) whose common core is a more or less
radical break with some traditional tenets of cognitivism. Unfortunately, even after
three decades the debate is still fledgling and there is no even remotely unanimous
usage of key notions like ‘embodied’, ‘embedded’, ‘extended’, ‘distributed’, or ‘enacted’.
This section provides the outlines of a conceptual clarification.
Since the departure from cognitivism is said to consist in the insight that cognition
involves not only intracranial, but also extracranial processes, there are two crucial
questions: “what does it mean to say that cognition involves extracranial processes?”
and “what does it mean to say that cognition involves extracranial processes?”4
The first question draws attention to a relational dimension: how are cognitive
processes on the one hand and whatever processes have to be invoked over and above
intracranial ones on the other hand related? While some insist that cognitive processes
are constituted by extracranial processes, others argue that they non-constitutionally
depend upon extracranial processes, where constitution is understood along the lines
of the mereological part-whole relationship and dependence either causally or
evolutionarily (in the sense that something has been designed to function only in
combination with something else on which it depends). Intuitively, a computer is
constituted (in part) by the CPU and the RAM, while the power plant producing the
electricity is not among its constituents, but only something upon which its
functioning causally depends.
The second question draws attention to a locational dimension: what kinds of
processes co-determine or co-constitute cognitive processing, over and above the
intracranial ones? While some have focused on the body’s contribution to cognitive
processing (where “body” must be read as “body minus brain,” because otherwise
the fact that the brain is part of the body would make the position collapse into
 Philosophical Psychology 69
cognitivism; see note 4), others have focused on the extrabodily environment’s
contribution.
This leads to four different situated hypotheses that vary in a twofold way along the
relational and locational dimension: cognitive processing may be (1) co-dependent
upon bodily processes, (2) co-constituted by bodily processes, (3) co-dependent upon
extrabodily processes, or (4) co-constituted by extrabodily processes. (1) and (2)
capture the idea that cognition is embodied, (3) the idea that it is embedded, and (4) the
idea that it is extended.
In addition, enactivists characterize cognition as an active engagement in which a
cognitive system brings about (“enacts”) its environment (Umwelt), thereby rejecting
the locational question. For them, it is at best misleading and at worst senseless to ask
whether cognitive processing takes place in the brain, in the body, in the environment,
or in any combination of these, given that cognition is supposed to be an essentially
relational and temporally extended process of “sense-making” (Thompson &
Stapleton, 2009) by an autonomous and adaptive system in interaction with its
environment, so that to ask where a particular cognitive process is taking place at a
particular time is to miss the very explanans.
Finally, cognitive processes have been said to be distributed over complexes of
interacting agents and technological resources, for example when the navigation of a
vessel is accomplished by the concerted effort of the technologically equipped crew on
the bridge (e.g., Hutchins, 1995). To avoid conceptual muddle, one should distinguish
cognitive distribution and cognitive extension. Cognitive extension usually concerns a
single system or agent at the “center” of cognitive processes, from where they are then
said to extend (under certain conditions) beyond the original organismic boundaries,
leading to an extended cognitive system. The point of distributed cognition, in
contrast, is that cognitive processes are “spread out” over a collective, no individual
member of which can be singled out as the “center” of these processes.5
Equipped with these rather pithy classifications, let us return to affectivity. Section 4
draws attention to an important difference between cognition and emotion which
makes the environment’s contribution to our affective life much more interesting than
the body’s contribution. The remaining sections will then be concerned with different
kinds of environmental contributions.

4. Embodied Cognition and Embodied Emotions

When the idea that the specific morphological, biological, and physiological details of
an agent’s embodiment make a special and ineliminable contribution to its cognitive
life gained prominence in the last quarter of the twentieth century, it stood in stark
contrast to the received view of the time, which regarded cognition as a rather abstract
and detached process of representation transformation. In contrast, the claim that the
body substantially contributes to an agent’s affective life should have been
uncontroversial from the very start. Emotions have, except perhaps for a short
period of radical cognitivism, never been taken to be a purely abstract, “fleshless” affair
70 A. Stephan et al.
to begin with (section 2). Ever since Aristotle pointed out that anger might be
considered the striving for revenge by the dialectician but the boiling of the blood
around the heart by the natural scientist (De anima, 403a/b), the body has played a
pivotal role in almost all accounts of emotions. Even a staunch cognitivist like
Solomon eventually conceded that in pursuit of an alternative to feeling theories he
“had veered too far in the other direction,” acknowledging that “accounting for the
bodily feelings . . . in emotion is not a secondary concern and not independent of
appreciating the essential role of the body in emotional experience” (2004, p. 85).
The mere claim that the body makes a special and ineliminable contribution to our
affective life can thus not be one of the hallmarks of a new generation of situated
approaches to emotion—we knew that all along. In this sense, the claim that emotions
are embodied is less interesting than the claim that cognition is embodied.
It is not entirely uninteresting though. First, quite generally, the question whether
(some aspects of) emotions are co-dependent or co-constituted by bodily processes is
an important issue in the philosophy of emotions. Second, in light of component
theories of emotions (section 2), the question is not whether the body contributes to
our affective life per se, say in form of bodily arousal or facial expression, but whether
it also contributes to what is generally agreed to be purely cognitive, namely, the
appraisal component. If it turned out that the body also contributes to our cognitive
appraisal of, say, a colleague’s offense as annoying, anger would be embodied in a
hitherto unappreciated way (Colombetti, 2007, pp. 536– 538). But in order to show
that, it is not sufficient to simply call appraisals “embodied,” without showing exactly
how the body contributes to what is traditionally considered to be the purely cognitive
appraisal. Prinz (2004), for example, talks about “embodied appraisals,” but for him
the bodily responses merely draw our attention to events in the world, while the
latter’s relevance, say as annoying, is in fact assessed only by so-called “elicitation
files”—which are in the brain. Clearly, for appraisals to be embodied, more is required
than that (note 4).
Things are different with regard to the role of the environment. While emotions are
typically conceived of as responses to changes in the environment, there is no pre­
established consensus that over and above that the environment has a more substantial
impact on our affective life. Detailing the exact way(s) in which the environment
substantially contributes to our affective life will thus further our understanding of
emotions, regardless of whether the environmental influence concerns the cognitive,
the bodily, or the experiential aspect. Let us therefore concentrate on the contribution
of the environment and our interaction with it (for a detailed discussion of embodied
emotions, see Wilutzky, Stephan, & Walter, 2011).

5. Embedded Emotions
Not any extracranial influence on our affective life renders emotions “situated.” A slim
or buff body may enhance our emotional well-being, and a prolonged illness or bodily
fatigue may lead to a major depression, but this does not mean that the relevant
 Philosophical Psychology 71
emotions are “embodied” in any interesting sense, for these bodily factors are mere
triggers, or elicitors, of emotional states. Likewise, not every environmental influence
on an emotion makes the latter “embedded.” The structure and character of the
environment can undoubtedly influence our affective life, for example when one is
disgusted by a fascist rally, or when a usually reserved person is carried away by the
euphoria of a cheering crowd. But it would trivialize the idea of situated affectivity to
call the corresponding emotions “embedded,” for no one has ever denied that the
environment contributes to our affective life in the sense that emotions are responses
to changes in the environment that are of import to us. What is needed is a non-trivial
notion of “embeddedness” that distinguishes cases where the environment is a mere
trigger from those in which it contributes to emotions in a way incompatible with
traditional accounts.
This problem is familiar from the debate about situated cognition. Advocates of
embedded cognition have to show in which sense they go beyond traditional
cognitivism, given that their claim that cognitive processes depend upon extrabodily
processes is entirely compatible with the cognitivist’s view that they are intracranial,
syntactically driven operations on representations.6 Two ideas have been invoked in
response. First, the partial dispensability of internal representations, and second, the
use of an appropriately structured environment or the active structuring of the
environment with the goal of reducing cognitive load, the so-called “scaffolding”
(e.g., Clark, 1997, p. 63). While cognitivism may be compatible with an environmental
dependence per se, it clearly seems to be incompatible with the specific kind of
environmental dependence envisaged by the advocate of embedded cognition. For the
cognitivist, cognitive processing depends upon the environment in the purely
counterfactual sense that if the environment were different, then the internal
representations of the environment were different, and if these representations were
different, then the internal cognitive processes were different. In contrast, advocates of
embedded cognition take the environmental dependence to be immediate and active.
The invocation of the structured environment as an external scaffold makes any
mediation by elaborate internal representations (partially) dispensable because it
replaces (or at least augments) the intracranial transformation of passively received
representations by the active manipulation of the relevant external structures
themselves. Embedded approaches to cognition are thus incompatible with
cognitivism not because of the environmental dependence per se, but because the
kind of dependence posited by the former is unavailable to the latter.
A rather similar picture emerges in the case of emotions. Regarding the first point,
the role of the environment as a scaffold, examples where we actively structure or use
the appropriately structured environment as an “affective scaffold” in order to
influence our emotional well-being abound. We furnish our apartment in such a way
that we feel comfortable, we remove everything that reminds us of our ex-partner to
alleviate the pain of separation, we deliberately undergo a therapy in order to get over
our anxieties, etc. The idea of active structuring is in fact crucial for strategies of
emotion regulation (e.g., Gross, 2002). It makes a difference for one’s emotional life
whether one decides on the eve of an important exam to meet with one’s equally
72 A. Stephan et al.
nervous classmates or an old friend (“situation selection”), and if one goes to see one’s
friend, it makes a difference whether one discusses the imminent exam and the
consequences of failing or just schmoozes and indulges in good memories from the
common past (“situation modification”; see Stephan, 2012, in particular section 3).
Griffiths and Scarantino (2009) also stress the role of the environment as an actively
structured scaffold for our emotional engagement with the world. They reject both
purely cognitivist accounts and neo-Jamesian “embodiment”-theories (e.g., Prinz,
2004), which in their eyes denigrate the environment to the mere location of the input
to and the output of emotional responses (Griffiths & Scarantino, 2009, p. 437).
Instead, they conceive of emotions as a form of skillful engagement with the world that
can be scaffolded, both synchronically in the unfolding of a particular emotional
performance and diachronically in the acquisition of an emotional repertoire
(Griffiths & Scarantino, 2009, p. 443). The construction of sacred buildings with a
religious function, for example, is aimed at providing specific atmospheres
that support religious feelings of sublimity or humility (e.g., Anderson, 2009).
In particular, the provision of confessionals in churches enables certain kinds of
emotional performance (synchronic scaffolding), and the broader Catholic culture
supports the development of the ability to engage in the emotional engagements of
confession (diachronic scaffolding). In a similar vein, social emotions like love and
guilt function as commitment devices (e.g., Frank, 1988) that can be (partially)
offloaded onto social institutions. In the case of romantic love, for instance, marriage
creates social pressure that helps ensure mutual support and fidelity, the tradition of
exchanging gifts at wedding anniversaries acts as an affective scaffold by making sure
the everyday routines of married life are regularly interrupted by acts of romantic
generosity, etc.
Regarding the second point, the (partial) dispensability of elaborate internal
representations, Griffiths and Scarantino’s account is also illuminating. On their view,
emotional content need not have a conceptual representational format; it may instead
have a “fundamentally pragmatic dimension, in the sense that the environment is
represented in terms of what it affords to the emoter in the way of skillful engagement
with it” (2009, p. 441). This supposedly holds even for so-called “higher” cognitive
emotions like shame, guilt, or embarrassment. Authentically expressed guilt, for
instance, can be understood as a social strategy (see, again, Frank, 1988) aimed at
reconciliation in order to repair a relationship, and thus as a form of a skillful
engagement for which the conceptual capacities and elaborate internal representations
presupposed by classical cognitivist emotion theories are less important than has
traditionally been assumed (Wilutzky & Stephan, forthcoming)—as is evidenced by
the fact that social “guilt strategies” can already be observed in infants that do not (yet)
have the conceptual and representational repertoire required by traditional accounts
(e.g., Reedy, 2000).
If the ideas just sketched are on the right track, the two key ideas from the debate
about embedded cognition—the partial dispensability of internal representations and
the use of the (actively) structured environment as a scaffold—turn out to have close
analogues in the affective realm, in which case at least some emotions turn out to be
 Philosophical Psychology 73
embedded in a substantial sense not captured by extant accounts in the philosophy of
emotions that have hitherto focused too narrowly on affective phenomena of a certain
individualistic kind. Some emotions are dynamic engagements with the world rather
than inner snapshot responses to external triggers.

6. Emotions beyond Brain and Body

Interesting as this idea of embedded emotions may be, it does not address the
questions with which we began: do some emotions cross an individual’s boundary?
If so, what kinds of supra-individual systems can be said to have affective states? And,
crucially: why? For even if the environment is a (potentially indispensable) scaffold for
an individual’s affective life that renders elaborate internal representations
dispensable, embedded emotions do not cross organismic boundaries. In order to
show that emotions indeed transcend the bounds of brain and body, an additional
argument is needed. To see this, consider the situated approach of Slaby, who
discusses emotional episodes much like the ones described above (e.g., being drawn
into euphoria on an exuberant party) and argues that in such cases “a part of the world
is what sets up, drives, and energizes our emotional experience” (forthcoming, p. 9),
concluding that the environment provides “‘tools for feeling’ . . . in something
like the way there are ‘tools for thinking’ in EM [extended mind] theorizing”
(forthcoming, p. 10). Without any justification for the transition from a dependence­
to a constitution-claim, however, this is just an instance of Adams and Aizawa’s (2008)
“coupling-constitution fallacy”: the mere fact that a part of the world sets up, drives,
and energizes our affective life obviously neither entails nor vindicates the claim that
emotions are partly constituted by that part of the world.
In the debate about extended cognition, the constitution-claim has been justified
either by appeal to a “parity principle” (PP), according to which extrabodily processes
are constituents proper if they play the same functional role as comparable internal
processes might play, or by appeal to the idea of an “integration by complementarity,”
according to which extrabodily processes are constituents proper if they complement
internal processes in a way creating hybrid systems with characteristics individuals
alone, de-coupled from their environmental resources, could not have (e.g., Menary,
2006). Let us consider the application of these two strategies to affective phenomena.
Ignoring a lot of details (for a more comprehensive discussion of PP and its
limitations, see Walter, 2010), PP holds that cognitive processes are partially
constituted by extrabodily processes if the latter play the same functional role as
comparable internal processes we would not hesitate to admit as constituents proper.
In the standard example of Otto the Alzheimer patient, for example (Clark &
Chalmers, 1998), the entries in Otto’s notebook are supposed to be part of what
realizes his memories because they play the same functional role vis-à-vis his cognitive
and behavioral competences as do biomemories in “ordinary” adults. Therefore, PP
provides a viable route to extended emotions just in case the internal constituents of
emotions have extrabodily functional equivalents. Hence, since most accounts take
74 A. Stephan et al.
emotions to have various aspects (section 2), much depends upon what internal
constituent one is talking about. According to Scherer’s (2005) “component process
model,” for example, emotions are constituted by five highly interacting components:
(1) a subjective feeling component (experiences); (2) a cognitive component
(appraisals); (3) a motivational component (action-tendencies); (4) a neurophysio­
logical component (e.g., bodily symptoms); and (5) a motor expression component
(e.g., mimics). Barring any other constituent of emotions, one of these must have
extrabodily functional equivalents, if PP is to support the idea of extended emotions.7
Since PP requires that the internal component is exhausted by its functional role
(otherwise functional equivalence does not guarantee parity), while experiences
notoriously resist a functional characterization, PP is inapplicable to the subjective
feeling component.8
Since cognitive states or processes do seem susceptible to a functional analysis, the
appraisal component fares better in this regard. Nevertheless, it is not obvious that
extrabodily processes actually play the functional role characteristic of internal
appraisals: Otto’s notebook entries are at best dispositional, or standing, beliefs, while
appraisals resemble occurrent beliefs that are triggered in concrete situations by
concrete events and must be capable of influencing an agent’s behavior in timescales of
milliseconds. At least currently existing extrabodily resources apparently fail to be
functionally equivalent in this regard.9 However, much depends upon whether a full-
fledged equivalence is required or whether a coarse-grained equivalence along the lines
of commonsense functionalism is enough (e.g., Clark, 2008, p. 88). If not every fine-
grained detail regarding the rapid, automatic, and unconscious integration of the
information provided by the appraisal needs to be replicated,10 then appropriately
connected devices that inform an agent about threatening or undesirable situations
(say, radioactive pollution or speed traps) will be candidates for extended appraisals.
Although the information provided by such devices can of course not be integrated
with the other emotion components as intricately and dynamically as the information
in “normal” appraisal processes, the non-automatic and conscious evaluations in
question can nevertheless affect the agent’s experiences and action tendencies.
Or consider Otto’s cousin Arnold, an autistic person incapable of directly perceiving
and recognizing the emotional states of others in social interactions. If Arnold is
equipped with a headset camera connected to a computer running a program for
decoding human emotional states, his appraisal system is supplied with online-
information in real time about the emotional states of his interaction partners
(el Kaliouby, Picard, & Baron-Cohen, 2006). On the basis of this, Arnold can
immediately appraise the situation, allowing him to adequately interact with other
people. The information provided by such devices may not be processed as quickly in
Arnold’s appraisal system as in “normal” appraisal processes. As indicated above,
however, the question is whether such fine-grained differences matter or whether PP
requires only a coarse-grained functional equivalence, so that information provided
and integrated within time spans that allow for conscious and non-automatic appraisal
processes and inferences could still be functionally equivalent. One reason for thinking
that a coarse-grained equivalence suffices is the following: the emotional responses of
 Philosophical Psychology 75
high-functioning Asperger patients are slowed down considerably because they have to
deliberately and consciously draw inferences about the emotional states of others. But
even if these slow, non-automatic, and conscious inferences were the only basis of their
emotional responses, we would not take that to be a reason for denying them whatever
emotional competences they have (just as we take the amazing cognitive capacities of
autistic savants as evidence that they have a quite unusual kind of memory, rather than
concluding from the fine-grained functional peculiarities of their mnemonic skills that
they do not possess any memory at all; see Kyselo & Walter, 2011).
If only a coarse-grained functional equivalence is required, then PP is applicable to
the motivational component of emotions as well. Arnold’s device may not only inform
him about the emotional states of others, but directly elicit action-tendencies by
suggesting, say, when it is a good time to interrupt a conversation or when it is
appropriate to apologize for interrupting, etc. (although these action-tendencies may
again not be integrated with the other components as rapidly, automatically, and
unconsciously as in “normal” cases).
In the case of neurophysiological components (an increase in blood pressure, say, or
heightened muscle tone) a device functionally equivalent to the entire human body’s
physiology is admittedly hard to conceive. However, one may isolate smaller
functional units of the neurophysiological component for which functional
equivalents can be established, such as hormone secretion: a small device containing
vials of hormones known to play a central role in emotion processes (e.g., adrenaline,
oxytocin) may be activated by the appraisal component and as a consequence release
controlled amounts of hormones into the blood stream. Likewise, an apparatus similar
to a cardiac pacemaker that can increase or slow down the heartbeat in accordance
with the appraisal component’s activity could parallel the modulation of this singular
physiological response in normal occurrences of emotion.
If one is willing to countenance bodily expressions as parts of emotions,11 motor
expression components seem to be other candidates for an extension via PP. Instead of
grinding one’s teeth in anger or smiling in excitement, one may throw a plate against
the wall or fire celebratory gunshots in the air. Fine-grained functional equivalences,
however, may be unattainable, here, too, for bodily expressions like grinding our teeth
or smiling seem to be much more involuntary and much less under the agent’s
conscious control than their alleged extrabodily counterparts.
Although more work is obviously required here, it seems fair to conclude that if one
can justify the appeal to coarse-grained functional equivalences along the lines above
and resolve the other problems with PP well-documented in the debate about
extended cognition, then at least some emotions can (in a narrow range of special
cases) be said to be extended on the basis of PP.
What about the idea of integration by complementarity that for example Slaby
favors? The main problem here is that the appeal to complementarity alone does
nothing to justify the move from dependence- to constitution-claims and therefore
fails to provide an argument for cognitive extension. Not any external resource that
enables individuals to do something they could not do otherwise (e.g., cognitive tools
that come with representational formats, processing speeds, or bandwidth capacities
76 A. Stephan et al.
unavailable to internal resources on their own) is thereby ipso facto an extrabodily
part of their cognitive machinery. Unless we want to commit another coupling-
constitution fallacy, the mere fact that, say, we could not see without light should not
make the light rays radiating from the sun constituents of our visual perception—they
complement our cognitive machinery, but they are not (on pain of so-called “cognitive
bloat”) a part of it. Without any additional criterion that distinguishes the
“interesting” couplings that give rise to integrated hybrid systems and real extension
from those “mere” couplings where external resources complement the cognitive
machinery of a system without literally becoming a part of it, the integrationist’s
strategy therefore does not get off the ground. PP would obviously be one candidate
for such a criterion: the “interesting” couplings could be exactly those where the
extrabodily resource is such that were it internal, we would have no hesitation in
recognizing it as a part of the cognitive or affective process. That, however, would
make integrationism dependent upon PP, to which it was supposed to be an
alternative.12 Of course, the integrationist can always propose another criterion (or
accept a radical cognitive inflation, which, we venture, would amount to a reductio of
integrationism), but so far no convincing candidates are in sight, and, as said above,
barring any such criterion, the appeal to integration by complementarity alone cannot
close the gap between embedded and extended emotions.13
Although unsuitable as an argument for cognitive extension, the idea of integration
by complementarity can draw our attention to some interesting phenomena in which
an individual’s affective life is enriched by the coupling with other individuals in such
a way that complex supra-individual systems are capable of feats that do not come into
view as long as one merely concentrates on interacting individuals qua individuals.
Consider, for example, Scheler’s (1923/1954) category of feeling with (Miteinander­
fühlen) in the sense of collective affectivity. Discussing an example of mother and father
mourning at their child’s grave, Scheler argues that the interesting point is not that
each of the two is suffering, that each knows the other is suffering or that each is
suffering with the other, but that they are suffering together in the sense that they, as a
collective, share (non-metaphorically) the same pain (Leid). In cases like these,
suffering is not restricted to the organismic boundaries of the father or the mother: the
mother’s suffering may of course provide the input to the father’s emotional state, and
the father’s suffering may perhaps be dependent upon the mother’s suffering and the
rest of the environment in some more substantial sense. But over and above that, there
is also the suffering of the two together that does not come into view if one focuses
only on individuals as the bearers of affective states.
Apart from Scheler’s potentially controversial idea of inter-individually shared
experiences, more mundane examples of “joint emotions” include social interactions
in which emotions are dynamically coupled with a social environment that is not only
influenced by but also influencing the unfolding of individual emotional episodes.
In such cases, there is not just one single emotional reaction to someone else’s action,
but a continuous exchange between socially interacting agents; the affective
phenomena in question are not one-shot affective responses to a detected stimulus
(as in classical examples such as the lone hunter in the Savanna who comes across a
 Philosophical Psychology 77
lion and shows a fear reaction), but must be seen as dynamically unfolding between
social agents, where the outcome is initially open, with many factors influencing the
development of this process, such as the social setting, cultural conventions, and
practices. In such interactions, affective signals are sent back and forth, are received by
either party and shape the emotional responses on-line as, e.g., in a marital argument
(Griffiths & Scarantino, 2009, in particular p. 438; Wilutzky & Stephan, forthcoming).
Dynamic affective phenomena of this kind are not a mere philosophical fancy and by
no means unusual, but a well-documented research topic in other disciplines as well.
Cole (2009), for example, provides an interesting and illuminating psychiatric
perspective on the role of social interaction for emotional experience, and Goodwin
and Goodwin (2000) view emotions from an anthropological perspective as social
phenomena organized and made visible through situated practices used by individuals
to construct their lifeworld.
Other examples of “collective affects” that have their place “between” and not “in”
social beings include what Anderson (2009) calls “affective atmospheres” (Stephan,
2012, section 2). Such atmospheres can be encountered in particular spaces of time,
landscapes, and buildings, for instance as the mirthless atmosphere of a rainy
November dusk, the magnificence of the starring sky in the Andes, or the holy aura of
an old cathedral. Typically, such atmospheres are also present in certain epochs, say as
a revolutionary atmosphere, but they also ubiquitously emerge in social encounters,
for example when a group produces an icy or very welcoming atmosphere for a
person, say, in a job interview, and can have stable characters, nearly as objective as
secondary qualities.
Cases like these, where supra-individual systems are not (like, say, Arnold)
composed of an individual coupled to some technical or non-technical artifact or
natural resource but by groups of interacting individuals are the best candidates for
affective phenomena that cross an individual’s boundaries. PP or the idea of
integration by complementarity play no argumentative role here; the motivation for
not restricting affective phenomena of this kind to individuals (or their brains) is that
they are an essentially collective and emergent affair. First, just as a vessel is not
navigated by the navigator using his shipmates as an extrabodily resource, but by the
navigational crew as a whole, and just as a bill is not passed by one senator relying on
the other senators as extrabodily resources, but by the senate as whole, the oppressive
atmosphere that emerges when a job interview doesn’t go well and becomes awkward
or the contagious euphoria or panic of a crowd do not have a single individual as a
bearer but are distributed over supra-individual collectives of interacting individuals.
Second, while they supervene upon the individual states of the group’s members plus
the nonsocial environment, they are not just the aggregate of individual affective states
(as when, say, the three of us are a happy crowd in the sense that each of us considered
individually is happy). Rather, they are emergent in the sense that the affective states
and actions of each individual member continuously and reciprocally influence each
other and are themselves shaped and amplified in a top-down manner by the overall
dynamics of the group as a whole—a characteristic and psychologically well-
documented feature of emotional episodes in groups, in which otherwise sensible and
78 A. Stephan et al.
peaceful people can sometimes get carried away by the contagious nature of euphoria,
panic, or the blind rage of lynch mobs as a consequence of “deindividuation.”14
Strictly speaking, however, emotions of this kind are not “extended,” but
“distributed” (section 3). What is characteristic for them is social groups that are
collective subjects of emotions that emerge from the mutual interaction of their
members and for which no single member of the group is an appropriate bearer, not
individuals whose emotions are in part realized by some extrabodily process. Extended
emotions of the latter kind (those that e.g., Arnold might be said to have) are (still)
rather exceptional, requiring the special circumstances in which PP is applicable. The
more familiar cases we have in mind are not ones where an individual’s emotions are
extended into extrabodily, maybe even social, resources (as would be the case if, say,
Arnold instead of using his technological device instructed his wife to provide him
with information about the emotional states of others), but ones where emotions
are distributed over all the members of a group (as in the case of joint emotions),
or emerge from the interactions within a social group and are “out there” to be felt
(as in the case of atmospheres).15 Stressing the essentially dynamical nature, one may
call affective phenomena of this kind “enacted” rather than “distributed.” It should be
clear, though, that by doing so one does not (and should not) commit oneself to the
more controversial claims of the enactivist (section 3; see also Colombetti, 2007;
Colombetti & Thompson, 2007), and in particular not to the claim that emotions, due
to their essentially relational nature, have no location. They do. It’s just that some
emotions sometimes have their place beyond the brain and the body.

Acknowledgements
The authors are grateful to the Volkswagen Foundation who supported their work by
grant II/84 051.

Notes
[1] Compare, for instance, Marr’s (1982) purely cognitivist account of vision with
contemporary situated approaches like Noë’s (2004). While others use “situated”
synonymous with what we call “embedded” (Shapiro, 2010) or “extended” (Wilson &
Clark, 2009), we use it as an umbrella term for any departure from cognitivism stressing the
importance of the body, the environment, and/or the interaction of brain, body, and
environment (Robbins & Aydede, 2009).
[2] Or sense-feel-appraise-act-, sense-appraise-act-feel-, or sense-act-feel-cycles, depending on
your favorite account of emotions.
[3] A crucial question for componential theories of emotion is whether, if cognition is situated,
such a neat componential analysis of emotions with distinctly separable bodily and
cognitive components is at all possible (Colombetti, 2007).
[4] Some restrict the departure from traditional cognitivism to the fact that allegedly central,
amodal cognitive processes involve (pace, e.g., Fodor, 1983) sensorimotor areas of the brain.
However, such a notion of situatedness is entirely compatible with ascribing situated
cognitive processes to, say, brains in vats, and thus too weak to be worth that name.
 Philosophical Psychology 79
[5] While Clark and Chalmers (1998) already hint at the possibility of a “socially extended
mind,” where, for example, memories are shared between multiple agents, this nomenclature
obscures this important difference: the very talk of “extended” cognition suggests that there
is an individual whose cognitive processes ex-tend into the environment. In transactive
memory systems (e.g., Barnier, Sutton, Harris, & Wilson, 2008), however, or cases like the
joint navigation of a vessel, there is no such original bearer. The navigational processes do
not ex-tend from, say, the navigator into his fellow shipmates in the same way as, say, Otto’s
memories (see below) extend into his notebook; since their bearer is the distributed system
to begin with, there is simply nothing from which they can ex-tend. This distinction between
extended and distributed processes (see also Hutchins’ contribution to this issue) will allow
us in section 6 to distinguish different kinds of supra-individual affective phenomena. We
are indebted to an anonymous referee for pressing us to clarify this issue.
[6] The same problem arises for those who explicate the idea that cognition is embodied by
saying that cognitive processes depend upon bodily processes.
[7] Slaby rejects the appeal to PP on the grounds that “it is impossible to assign emotions clear-
cut functional roles that capture all their relevant aspects” (forthcoming, p. 12). This is true,
but no argument against PP, given that emotions can be extended even if not all their
relevant aspects are extended.
[8] Another reason for restricting experiences to the brain can be found in Clark (2009).
[9] Since the opponents of situated approaches admit that cognition (or affectivity) might be
extended (e.g., Adams & Aizawa, 2008), the only interesting issue is whether they actually
are extended.
[10] With regard to cognition, it has been argued that such fine-grained details do matter and
that the states of the Otto-cum-notebook system fail to count as beliefs or memories
properly so called because they do not exhibit the kind of rapid, automatic, and unconscious
informational integration characteristic of beliefs (Weiskopf, 2008), or the kind of recency-,
primacy-, and chunking-effects characteristic of memories (Adams & Aizawa, 2008, p. 61);
for a dissenting view, see Kyselo and Walter (2011).
[11] One of us (S.W.) is highly skeptical: to say that someone blushed because she was ashamed is
to offer a causal explanation of her blushing, not a mereological explanation that accounts
for the presence of a part in terms of the presence of the whole.
[12] As an anonymous referee has pointed out, it is not even clear that the appeal to PP and
integrationism are compatible, for given the different format and dynamics of the external
resources, functional equivalence may be unattainable. Yet, much depends again upon how
fine-grained the functional equivalence has to be (see above).
[13] Slaby (forthcoming) fails to see this, as does Menary (2006) in the debate about extended
cognition.
[14] For a meta-analysis of the psychological literature on deindividuation, see Postmes and
Spears (1998).
[15] Huebner’s (2011) notion of “genuinely collective emotions” bears some resemblance to this
idea, although he, erroneously in our eyes, appeals to parity considerations. According to
Wilson’s (2005) social manifestation thesis, collectives are unsuitable as the bearer of
cognitive (or, in our case, affective) processes, even though some individualistic cognitive
(or affective) processes require their bearer to be a member of an appropriately organized
and technologically equipped social collective. But as said above, a bill is not passed by one
senator embedded in a collective of senators, but by the senate as whole (which, as a whole,
is also responsible for passing it). Likewise, while it is true that the blind rage of a lynch mob
or a mass riot is possible only in the context of the mob or the mass, as Wilson’s social
manifestation thesis holds, the rage itself, emerging as it does from the dynamical, top-down
influenced interplay among the members, is clearly not a feature of any single member, but
of the collective.
80 A. Stephan et al.
References
Adams, F., & Aizawa, K. (2008). The bounds of cognition. Malden, MA: Blackwell.
Anderson, B. (2009). Affective atmospheres. Emotions, Space and Society, 2, 77 – 81.
Barnier, A., Sutton, J., Harris, C., & Wilson, R. (2008). A conceptual and empirical framework for the
social distribution of cognition. Cognitive Systems Research, 9, 33– 51.
Clark, A. (1997). Being there. Cambridge, MA: MIT Press.
Clark, A. (2008). Supersizing the mind. Oxford: Oxford University Press.
Clark, A. (2009). Spreading the joy? Mind, 118, 963– 993.
Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58, 7 – 19.
Cole, J. (2009). Impaired embodiment and intersubjectivity. Phenomenology and the Cognitive
Sciences, 8, 343– 360.
Colombetti, G. (2007). Enactive appraisal. Phenomenology and the Cognitive Sciences, 6, 527 – 546.
Colombetti, G., & Thompson, E. (2007). The feeling body. In W. Overton, U. Muller, & J. Newman
(Eds.), Developmental perspectives on embodiment and consciousness (pp. 45– 68). New York:
Erlbaum.
Dai, D., & Sternberg, R. (Eds.). (2004). Motivation, emotion, and cognition. Mahwah, NJ: Erlbaum.
Dennett, D. (2000). Making tools for thinking. In D. Sperber (Ed.), Metarepresentations (pp. 17– 30).
Oxford: Oxford University Press.
el Kaliouby, R., Picard, R., & Baron-Cohen, S. (2006). Affective computing and autism. Annals of the
New York Academy of Sciences, 1093, 228– 248.
Fodor, J. (1983). The modularity of mind. Cambridge, MA: MIT Press.
Frank, R. (1988). Passions within reason. New York: Norton.
Goldie, P. (2000). The emotions. Oxford: Oxford University Press.
Goodwin, M., & Goodwin, C. (2000). Emotion within situated activity. In A. Duranti (Ed.),
Linguistic anthropology (pp. 239– 257). Malden, MA: Blackwell.
Griffiths, P., & Scarantino, A. (2009). Emotions in the wild. In P. Robbins & M. Aydede (Eds.), The
Cambridge handbook of situated cognition (pp. 437 – 453). Cambridge: Cambridge University
Press.
Gross, J. (2002). Emotion regulation. Psychophysiology, 39, 281– 291.
Huebner, B. (2011). Genuinely collective emotions. European Journal for the Philosophy of Science, 1,
89– 118.
Hutchins, E. (1995). Cognition in the wild. Cambridge, MA: MIT Press.
James, W. (1884). What is an emotion? Mind, 9, 188– 205.
Kyselo, M., & Walter, S. (2011). Belief integration in action. Philosophical Psychology, 24, 245– 260.
Marr, D. (1982). Vision. San Francisco, CA: W.H. Freeman.
Menary, R. (2006). Attacking the bounds of cognition. Philosophical Psychology, 19, 329– 344.
Noë, A. (2004). Action in perception. Cambridge, MA: MIT Press.
Nussbaum, M. (2001). Upheavals of thought. Cambridge: Cambridge University Press.
Postmes, T., & Spears, R. (1998). Deindividuation and antinormative behavior. Psychological
Bulletin, 123, 238– 259.
Prinz, J. (2004). Gut reactions. Oxford: Oxford University Press.
Reedy, V. (2000). Coyness in early infancy. Developmental Science, 3, 186– 192.
Robbins, P., & Aydede, M. (Eds.). (2009). The Cambridge handbook of situated cognition. Cambridge:
Cambridge University Press.
Scheler, M. (1954). The nature of sympathy (P. Heath, Trans.). London: Routledge. (Original work
published 1923).
Scherer, K. (2005). What are emotions? Social Science Information, 44, 695– 729.
Shapiro, L. (2010). James Bond and the barking dog. Philosophy of Science, 77, 400 – 418.
Slaby, J. (forthcoming). Emotions and the extended mind. In M. Sperber & C. van Scheve (Eds.),
Collective emotions. Oxford: Oxford University Press.
 Philosophical Psychology 81
Slaby, J., & Stephan, A. (2008). Affective intentionality and self-consciousness. Consciousness and
Cognition, 17, 506– 513.
Solomon, R. (1976). The passions: Emotions and the meaning of life. New York: Anchor Press.
Solomon, R. (2004). Emotions, thoughts, and feelings. In R. Solomon (Ed.), Thinking about feeling
(pp. 76– 88). Oxford: Oxford University Press.
Stephan, A. (2012). Emotions, existential feelings, and their regulation. Emotion Review, 4, 157– 162.
Thompson, E., & Stapleton, M. (2009). Making sense of sense-making. Topoi, 28, 23 – 30.
Walter, S. (2010). Cognitive extension. Synthese, 177, 285– 300.
Weiskopf, D. (2008). Patrolling the mind’s boundaries. Erkenntnis, 68, 265– 276.
Wilson, R. (2005). Collective memory, group minds, and the extended mind thesis. Cognitive
Processing, 6, 227– 236.
Wilson, R., & Clark, A. (2009). How to situate cognition. In P. Robbins & M. Aydede (Eds.), The
Cambridge handbook of situated cognition (pp. 55– 77). Cambridge: Cambridge University Press.
Wilutzky, W., & Stephan, A. (forthcoming). In search of the intentional objects and the adequacy of
emotional interactions. In S. Döring (Ed.), [title of book to be determined].
Wilutzky, W., Stephan, A., & Walter, S. (2011). Situierte Affektivität [Situated affectivity]. In J. Slaby,
A. Stephan, H. Walter, & S. Walter (Eds.), Affektive Intentionalität [Affective intentionality],
(pp. 283– 320). Paderborn: mentis.