Special issue: The language of emotion – conceptual and cultural issues

Numéro spécial: Le langage de l’émotion –

questions conceptuelles et culturelles

Julien A. Deonna & Fabrice Teroni

Taking affective explanations to heart

Abstract.  In this article, the authors examine and debate the categories of emotions, moods,
temperaments, character traits and sentiments. They define them and offer an account of the
relations that exist among the phenomena they cover. They argue that, whereas ascribing
character traits and sentiments (dispositions) is to ascribe a specific coherence and stability to
the emotions (episodes) the subject is likely to feel, ascribing temperaments (dispositions) is to
ascribe a certain stability to the subject’s moods (episodes). The rationale for this distinction,
the authors claim, lies in the fact that, whereas appeal to character traits or sentiments in
explanation is tantamount to making sense of a given behaviour in terms of an individual’s
specific evaluative perspective – as embodied in this individual’s emotional profile – appeal to
temperaments makes sense of it independently of any such evaluative perspective.

Key words.  Character trait – Emotion – Explanation – Mood – Philosophy – Sentiment –

Temperament

Résumé.  Dans cet article, les auteurs s’interrogent sur les catégories d’émotions,
d’humeurs, de tempéraments, de traits de caractère et de sentiments. Ils proposent des
définitions de ces catégories et offrent une analyse des relations existant entre les phénomènes
qu’elles désignent. Ils défendent ainsi l’idée selon laquelle attribuer des traits de caractère et
des sentiments (dispositions) revient à attribuer une certaine stabilité et cohérence dans les
émotions (épisodes) ressenties par le sujet, alors qu’attribuer des tempéraments (dispositions)
revient à attribuer une certaine stabilité dans ses humeurs (épisodes). Cette distinction est
motivée par les rôles explicatifs distincts joués par ces catégories. D’un côté, faire référence à
des traits de caractère ou à des sentiments dans l’explication psychologique, c’est faire sens
d’un comportement donné en termes de l’orientation évaluative d’un individu, en tant qu’elle
s’incarne dans son profil émotionnel. De l’autre, lorsque l’on fait appel à des tempéraments,
l’explication est indépendante d’une quelconque perspective évaluative.

© The Author(s), 2009. Reprints and permissions: http://www.sagepub.co.uk/journalsPermissions.nav

Social Science Information, 0539-0184; Vol. 48(3): 359–377; 106197
DOI: 10.1177/0539018409106197 http://ssi.sagepub.com
360   Social Science Information Vol 48 – no 3

Mots-clés.  Emotion – Explication – Humeur – Philosophie – Sentiment – Tempérament –

Trait de caractère

The variety of affect

Within the commonsense picture of our affective lives, the concepts of emo­
tion, mood, sentiment, character trait and temperament appear to play crucial
and distinctive explanatory roles. Yet, while we feel that there are differences
between these categories – explaining a piece of behaviour by reference to
sadness (emotion) is different from explaining it by reference to gloominess
(mood) or pessimism (character trait) – the truth is that the use of these terms
is far from systematic in ordinary language. Some of these terms are some­
times interchangeable, they are ambiguous and, more generally, whether a
given term denotes a character trait or a temperament, a sentiment or an emo­
tion is far from clear, except maybe to a few philosophers.
Regarding these categories and their relations, recent philosophy has con­
cerned itself mainly with one question: Does anything really correspond to
the category of character traits, and this independently of its manifest use in
folk psychological explanations?1 We shall be taking all these affective cate­
gories seriously – ignoring the recent and widespread scepticism vis-a-vis
character traits – and shall argue that, if they refer to anything at all, they
refer to distinct phenomena. This attitude, we believe, will shed light on the
structure and complexity of our psychological explanations. At the end of
our discussion, we will be in a position to trace the limits within which
scepticism about character traits is warranted. Before reaching this point,
however, our questions are as follows: What important distinctions should
be made within the affective domain? Which phenomena within this domain
are dispositional and which are not? What are the relations between these
phenomena?
We argue that emotions and moods are occurrent states that bear system­
atic and interesting relations to the other categories. Both ascriptions of
character traits and sentiments, on the one hand, and ascriptions of tempera­
ments, on the other, are ascriptions of dispositions. But the dispositions are
different. Whereas ascribing character traits and sentiments (dispositions)
points towards a specific coherence and stability in the emotions (episodes)
a subject is likely to feel, ascribing temperaments (dispositions) points
towards a certain stability in the moods (episodes) she is likely to feel.2 And,
as will become apparent, the rationale for this distinction lies in the fact that,
whereas appeal to character traits or sentiments in explanation is tantamount
Deonna & Teroni Taking affective explanations to heart   361

to making sense of a given behaviour in terms of a subject’s specific evalu­

ative outlooks and commitments, appeal to temperaments makes sense of
it independently of any values or commitments she might have. So, even
though psychological explanations in terms of sentiments, character traits
and temperaments give final answers to questions about the reasons why an
agent performed a given action, they do so in different ways.

Emotions and moods

Our discussion should naturally start with the two occurrent psychological
states that we claim are relevant to understanding the other categories of
affective phenomena.
Both emotions and moods are rightly viewed as occurrent states. That is,
both are marked by the fact that they take place at a given time, have a given
duration and are characterized by their phenomenological character.
Emotions and moods also motivate in typical ways. For instance, the anger
(emotion) John feels towards Maria might start just after she insults him and
last until she asks for forgiveness two minutes later. During this interval,
John has the experience of ‘what it is like’ to be angry and is likely to be
motivated to seek revenge. If Mike wakes up grumpy (mood), his grumpi­
ness might last until his first coffee. Here again, there is a distinctive way
Mike feels, one that philosophers often characterize by saying that moods
colour all the subject’s experiences during the relevant interval. And, as long
as Mike is grumpy, he will typically be motivated to act in various ways,
nervously roaming around his office, maybe, or banging doors.
How should we then distinguish emotions from moods? As has been often
stressed, the crucial distinction lies in their respective intentional structures.
While you are always angry at someone or something on this or that account,
your grumpiness is not directed at someone or something on this or that
account. Similarly, you rejoice over something and are sad about something,
but your joyful mood and your feeling down or depressed are not directed
towards anything in particular. This has to be understood as follows.
Emotions involve evaluations of specific objects, whereas moods do not.
For John to be angry at Maria’s remark requires him to evaluate this remark
as, say, offensive.3 On the other hand, Mike’s grumpiness does not require
him to take any evaluative attitude towards any specific object. This contrast
is made vivid by the fact that, while the idea that emotions are appropriate
or not makes perfect sense (Maria’s cheeky remark doesn’t make John’s
anger appropriate), it doesn’t make much sense to think of moods as appro­
priate or not.4
362   Social Science Information Vol 48 – no 3

Why is that so? One appealing explanation goes like this. John’s appre­
hension of Maria’s remark as offensive is for him a reason to be angry at her,
whereas nothing needs to be apprehended by Mike as a reason for his
grumpiness. Emotions are reason-responsive states, whereas it does not
make sense to speak in this way of reasons for moods. Of course moods
have, like emotions, causes (if not always clearly identifiable ones); Mike’s
grumpiness might be caused by his having a headache. Yet, while the head­
ache causally explains his grumpiness, it does not provide a rational or
irrational explanation for it. It has nothing to do with reasons. And this is
why changes in mood, as opposed to changes in emotion, are typically not
traceable to reasons. The reason responsiveness of emotions, a feature that
moods lack, explains in the case at hand why talk of appropriateness makes
sense with regard to the former but not with regard to the latter.
We have stressed how the respective types of explanations we give for the
occurrence of emotions and moods differ. Now, these differences reverber­
ate in the way we appeal to these states to explain behaviour. Suppose we
mention John’s anger at Maria over her remark as the explanation of his now
ignoring her. We explain his behaviour in terms of an affective state we con­
nect with a reason-responsive evaluation of its object, i.e. Maria’s offence.
Here a piece of behaviour is explained by a state characterized by an open­
ness to reasons and thus assessable in terms of (in)appropriateness. Suppose
we now refer to Mike’s grumpiness to explain the generally distracted and
even inconsiderate way he interacts with her. We here explain his behaviour
in terms of a state we consider as independent of an evaluation of Maria or of
any other feature of the situation: these properties play no role in explaining
why Mike behaves as he does. Rather, we refer to a non-reason-responsive
or an a-rational affective cause of his behaviour, which does nothing to jus­
tify it: we perceive moods as providing explanations of behaviour precisely
to the extent that properties of the situation are irrelevant in explaining it.
Thus, if both emotions and moods are motivational states, the pieces of
behaviour they elicit are subject to very different explanations. When we
feel a given emotion, the fact that we are ready to act in given ways is under­
stood, from the first- and from the third-person points of view, as a readiness
to act on a given object because we evaluate it in a given way. In anger, for
instance, I am ready to seek revenge because my anger evaluates a remark
as offensive. And this sharply contrasts with moods. This is of course closely
connected with the striking fact that we never mention moods to justify but
exclusively to excuse or to provide mitigating circumstances, whereas we
always perceive emotions as justified or not, and only some of the justified
ones occasionally serve as mitigating circumstances.
Deonna & Teroni Taking affective explanations to heart   363

This should not hide the fact that we also appeal to moods to provide
an apparently different type of explanation. Indeed, moods also feature in
explanations of states and pieces of behaviour that rest on evaluations of
objects or situations. For instance, Mike’s grumpiness is offered as an expla­
nation of his being angry at Maria in virtue of his taking her remark as
offensive. Here the mood is conceived as disposing subjects to apprehend
situations in a certain way. So, what does the mood add to the mere mention
of an emotion (anger) or a judgement (she is insulting) to explain Mike’s
behaviour? What is added is that the object and its properties that would
have made the emotion or judgement appropriate (for they are reason-
responsive states) do not play the relevant role in the subject’s evaluation,
which is wholly explained by the mood. As a result the explanation con­
ceives the evaluation as irrational. Thus, even when a piece of behaviour
rests on an evaluation of the relevant situation, reference to a mood implies
that the explanation differs in kind from the one that would be given by an
exclusive appeal to emotions or evaluative judgements.
So, although moods and emotions share a number of important traits, they
differ crucially: since moods are disconnected from any evaluative outlook,
we understand them as a-rational states.5 For this reason, when we appeal to
them to explain behaviour, the behaviour will qualify either as a-rational,
given its independence from any evaluation of the situation, or as irrational,
to the extent that moods give rise to evaluations whose sensitivity to reasons
they bias or even cancel.

Temperaments and moods

To continue our investigation, note that psychological explanations often do

not stop with the mention of moods. There are three main types of explana­
tion of why moods occur. The first proceeds in non-psychological terms,
since it refers to various physiological conditions, for example when we
explain grumpiness by a lack of caffeine. The second type refers to psycho­
logical states of some sort, typically affective states like intense emotions.
Mike’s grumpiness might be explained by the lingering anger he feels after
Maria asked for forgiveness. Indeed, even though emotions and moods dif­
fer, there are important dynamic relations between them: not only do intense
emotions often give rise to moods, but, as illustrated above, moods also
often elicit specific emotions.6 The third type, the one that concerns us here,
proceeds not by referring to previous mental episodes but to the subject’s
disposition to be in given kinds of moods. To revert to the example of Mike,
364   Social Science Information Vol 48 – no 3

we may explain his mood by saying something like ‘And, you know, Mike
is an irritable fellow’, i.e. we insert his mood in a recurring pattern of affec­
tive states of this nature – and here we point towards a stability in a subject’s
disposition to experience a given mood.
Although the relevant area of affective language is far from clear, we
venture to say that this third type of explanation is typically provided when
we use expressions such as ‘being irritable’ (the connection is here with a
disposition to be in a grumpy mood), ‘phlegmatic’ (being typically in calm
or relaxed moods), ‘melancholic’ (depressed or downcast), ‘impatient’ (fidg­
ety), ‘taciturn’ (gloomy) or ‘lascivious’ (aroused). These terms are typically
used to explain the occurrence of moods by reference to the subject’s ten­
dency to undergo such states. And, at least in these instances, these terms
refer to dispositions to enter into distinct internal climates or tempers,
exactly what the term ‘temperament’ in English and other languages appears
to capture.
If we take these considerations at face value, we may conclude that the
concept of a temperament is the concept of a disposition to recurrently expe­
rience moods of a certain kind. Once again, and to acknowledge the fuzzi­
ness of ordinary language in this area, it is not part of our claim that to each
mood term there corresponds a distinct temperament term, to each tempera­
ment term, a distinct mood term (e.g. it is not clear what mood term corre­
sponds to an energetic temperament or that we have more to say of a joyous
temperament than that she often is in a good mood), or that we always have
at our disposal two terms to distinguish the mood from its corresponding
temperament (e.g. we speak of a gloomy temperament as we speak of a
gloomy mood). Rather, our claim is that the language of temperaments is
typically used to include an additional layer of psychological explanation
that bears systematic and interesting connections with moods.
This conclusion can be further motivated if we consider the role of
temperament terms in explanation. For it bears striking similarities with expla­
nations in terms of moods. First, temperament terms are never used in
justification but serve to provide excuses or mitigating circumstances at best.
Offering Mike’s irritable temperament to explain his grumpiness and, through
it, a given piece of behaviour, does nothing to justify it. Second, explaining
behaviour by temperaments is very similar to explaining it by moods. On the
one hand, temperaments figure prominently in a-rational explanations of
behaviour. We refer to the subject’s irritable temperament, as we refer to his
grumpiness, to explain pieces of behaviour we perceive as detached from any
evaluation of the relevant situation. The only difference is that, whereas we
might want to know why (in the causal sense) a person might be grumpy, no
such question will arise once we refer to his being irritable. And in line with
Deonna & Teroni Taking affective explanations to heart   365

what we said about moods, we also conceive of temperaments as disposing

subjects to emotions or judgements that evaluate situations in specific ways
but that qualify as irrational because they are not properly responsive to these
situations’ properties. Third, and finally, the connections between tempera­
ments, moods and emotions are not conceived of in normative terms. We
expect the irritable person to be in given moods and, through them, to experi­
ence given emotions, but we do not perceive these moods and emotions as
something she should experience given her temperament.
These remarks motivate the claim that temperaments and moods are
related as dispositions to occurrences. Thinking of them in this way is not
only economical, it also explains the deep similarities between psychologi­
cal explanations in terms of temperaments and moods. So, when a mood is
explained in terms of a temperament rather than by circumstantial causes,
what is added is that the relevant piece of a-rational or irrational behaviour
is often, if not always, to be ascribed to a given mood.

Character traits and emotions

It is true that the language of character or personality traits,7 like the lan­
guage of temperaments, is far from systematic; and intuitions as to what
belongs to this category are less than steady. One approach would consist in
considering, à la Ryle (1949), all situation-dependent stability and consist­
ency in behavioural output as traceable to character traits. We believe that
this is the wrong approach to the subject matter. In this spirit, we argue for
a substantial thesis regarding character traits, whose full potential in illumi­
nating the complex structure of the affective domain becomes apparent as
we go along. The thesis has it, in line with a long tradition of thinking about
the virtues,8 that explaining a piece of behaviour by mentioning a character
trait serves to draw attention to the specificities of a subject’s emotional
profile in the explanation of how she behaves and reasons. This account
makes sense of at least these paradigmatic examples of character traits:
optimistic, kind, courteous, opportunist, meticulous, modest, boastful, loyal,
frivolous, cruel; but also negligent, insensitive, unfriendly; and of most of
what we conceive of as virtues and vices.
It is perhaps uncontroversial to say that the function of character traits
consists in connecting in a persistent and coherent way two dimensions of a
subject’s psychology. The first dimension is cognitive in nature. We conceive
of a subject to whom we attribute a character trait as having a distinctive
proclivity to apprehend situations in distinctive terms. In calling a person
kind, we characterize her as having a proclivity to apprehend situations in
366   Social Science Information Vol 48 – no 3

terms of the presence of others in need of help; in calling another one cruel,
we describe her as prone to evaluate situations in terms of opportunities to
make others suffer; and in describing someone as opportunistic, we claim
that he is prone to evaluate situations in terms that will further his interests.
Ascribing a character trait is thus to say that the subject is distinctively dis­
posed to apprehend situations in given terms, and this of course constitutes a
major difference with temperaments. The second dimension to which con­
cepts of character traits speak is motivational in nature. Thus, the kind person
is motivated to help those he apprehends as needing help; the cruel person is
motivated to make others suffer; the opportunist is motivated to act so as to
further his interests. These two dimensions are required for the ascription of
character traits. Thus neither a person with a proclivity to evaluate situations
in terms of others as being in need of help but lacking any motivation to help,
nor a person who is so motivated but never apprehends situations in the
relevant terms, will qualify as kind.
The fact that it appears possible for a subject to satisfy the cognitive
dimension of a character trait without satisfying its motivational dimension,
and vice-versa, is easily understood as implying that these two dimensions
are wholly independent from one another.9 Character traits, as they would be
conceived on the traditional ‘belief–desire model’ of psychological explana­
tion, indeed consist of two independent parts, a motivational part and a
cognitive one.
According to this model, the former consists in the presence of a generic
final desire.10 This desire, in the case of kindness, is the desire to help those
in need, a motive that generates specific desires in specific circumstances.
John’s generic desire to help those in need thus generates the specific desire
to help Sally in her dire family circumstances. The latter cognitive part of
character traits will have to do with the subject’s dispositions to form rele­
vant beliefs, i.e. John is disposed to notice that people’s predicaments are
weighing on them and in the circumstances thus forms the belief that Sally
is suffering. That is to say, when these beliefs connect with the generic
desire, they contribute to engendering the more specific desires. Note that,
because generic desires have nothing to do with the beliefs relevant for the
given character trait and, conversely, because these beliefs are independent
from the generic desires, the following will ensue. Kind and cruel individu­
als are distinctively disposed to notice that others are suffering, but their
beliefs will connect with very different generic desires, thereby engendering
distinct specific desires: the desire to help such and such a person in the
former case; the desire to aggravate her situation in the latter case. The
upshot is that, when we attribute a character trait, we may be in the business
Deonna & Teroni Taking affective explanations to heart   367

of attributing complexes of mental states, but the unity of these complexes

is a unity per accidens.
Should we be satisfied with the conclusion that character traits consist of
independent cognitive and motivational dimensions? As we indicated, we
believe that character traits exhibit a unity that the belief–desire picture does
not account for, and that only an appeal to emotions can provide the needed
unitary account. This is the alternative we now discuss.
The starting-point consists in arguing that the belief–desire picture rests
on an implausible assumption. As noted, according to this picture, people
with different character traits may share the same cognitive structure. For
instance, cruel and kind people are distinctively disposed to detect the pre­
sence of others in need of help, even though their motivations differ. But the
important intuition this picture cannot accommodate is of course that this
surface similarity hides deep psychological differences. This idea should be
pursued as follows.
The intuition is that kind and cruel individuals will reach the relevant
belief via distinctive sensitivities to situations that exemplify suffering. For
the kind person, the ‘mode of presentation’ of suffering will take a shape it
will not take for the cruel or the indifferent person. How should we qualify
this mode of presentation? It seems clear that it is affective through and
through. For instance, the presence of suffering will, for the kind person,
elicit sympathy; for her, the likelihood of the suffering of another will elicit
dread that it will really occur and the hope that it will not; if nothing is done
to relieve it, she will be likely to feel anger or indignation; if she can do
something about it and does, relief may ensue and if she does not, she will
likely feel guilt, remorse or shame, etc. And, of course, this affective struc­
ture is not shared by the cruel person, whose belief that others are suffering
will connect with wholly different emotions. Starting with the above intui­
tion, then, we should claim that a distinctive affective profile is associated
with each character trait. The crucial point to which we shall come back is
that character traits manifest themselves either straightforwardly in the
typical behaviour associated with the trait (helping, making suffer) – this
constitutes their stability – or, when for one reason or another they cannot
express themselves at a particular time or at all, they manifest themselves,
as we have just illustrated, in specific future-, present- or past-oriented emo­
tions; this makes for their coherence.
This account of character traits as specifications of a subject’s emotional
profile should now be elaborated in connection with their role in explana­
tions. To appeal to a character trait to explain a piece of behaviour is to refer
to the subject’s distinctive affective sensitivity. So the crucial question is to
368   Social Science Information Vol 48 – no 3

elucidate the difference between explaining a piece of behaviour by refer­

ence merely to an emotional episode as opposed to explaining it by refer­
ence to a character trait. The explanation of a piece of behaviour in terms of
an emotion, say compassion, proceeds in terms of the subject’s evaluating
the situation as exemplifying suffering that needs to be alleviated. This point
echoes what we said about emotional episodes at the outset, i.e. that distinct
emotions connect with distinct evaluations. And this is the case in the
absence of any character trait. Now, to say of an emotion and the behaviour
it motivates that they are grounded in a character trait is to specify the place
the type of evaluations implied by the emotion has in the subject’s life. To
claim that a piece of compassionate behaviour is motivated by kindness is
to claim that this emotional response is part of a larger pattern of real and
counterfactual evaluations the subject either does make, or would make, in
similar circumstances. It is to say that the response to suffering a given token
of compassion manifests is rooted in the person’s psychology in such a way
that suffering would produce such things as shame were it not acted upon or
indignation that others do nothing to alleviate it, etc. Character traits are now
understood in affective terms: they have to do with deeply rooted complexes
of dispositions to experience specific emotions in specific circumstances.11
In this sense, character traits specify the weight subjects lend to certain
values or disvalues.12
Two remarks are in order before we proceed. First, character traits can
specify emotional profiles either by pointing towards the presence of a sen­
sitivity to value, or towards its absence. Calling someone courteous or boast­
ful is to say that she is distinctively sensitive to a certain value, while calling
her insensitive or negligent is to say that she is blind to given values. In both
cases, a specification of the person’s emotional profile is offered. Second, it
should be stressed that a character trait will manifest itself in many mental
states besides emotions. The way the subject thinks (in an orderly fashion),
remembers (only the good times now gone) and imagines (that everyone is
after her) will often be the manifestation of a trait. Indeed, traits will often
manifest themselves in these ways; but within the present proposal this
should be explained through the specificities of given emotional profiles.
We can now appreciate why this alternative picture of character traits has
important consequences as regards their unity. We have claimed that a char­
acter trait is an emotional profile that fosters the apprehension of situations
in terms of the relevant value. And since the cognitive dimension of character
traits consists in dispositions to feel emotions, it can be claimed, in contrast
to the belief–desire model, that it is not mute as regards their motivational
dimension. For the occurrent emotions someone with a given character trait
will feel are not motivationally neutral. We mentioned at the beginning of our
Deonna & Teroni Taking affective explanations to heart   369

discussion the fact that types of emotions connect with distinctive types of
action readiness. This is of course not a brute fact about emotions but is
deeply connected to their cognitive structure: to experience an emotion is to
evaluate a given situation or object in a given way that prepares for action. In
fear and anger, one is respectively disposed to flee from a given situation or
to avenge oneself because one represents a situation as dangerous or as offen­
sive. Experiencing an emotion is not only evaluating a situation in a given
way, it is also to take a specific stance on it. And this means that, if the cogni­
tive dimension of character traits has to do with stable and coherent emo­
tional dispositions, this affective mode of apprehension consists eo ipso in the
subject’s taking a specific stance with respect to what is to be done in a given
situation. Because the emotions of the kind and the cruel person differ, say
compassion and Schadenfreude respectively, the cognitive evaluation they
embody and consequently their motivational impact will radically diverge.
Now, and this is crucial, although the cognitive and motivational dimen­
sions of character traits form within the present picture a unit that excludes
the possibility that the kind and the cruel person apprehend the world in the
same terms, this is not to say that their respective motivations will translate
straightforwardly into the typical behaviour associated with the trait. The
cruel person might also be opportunistic, or simply have had a wonderful
day, facts that might prevent him from manifesting his cruelty by cruel
behaviour. Traits operate neither in isolation from one another nor in isola­
tion from other motives, any of which might take, and often does take,
precedence over a given trait.13
This account of character traits in terms of sensitivities to values under­
stood as distinctive emotional profiles explains the distinct role they play, by
contrast with temperaments, in psychological explanations. It is true that we
conceive of temperaments and character traits as typically not something
we have for reasons. While we sometimes acquire character traits because
we have reflected on the importance of given values and trained ourselves
to be affectively sensitive to how they fare, this is the exception rather than
the rule. But this is the only relevant similarity between temperaments and
character traits, as we now argue. Character traits, we said, are dispositions
to experience emotions while temperaments are dispositions to experience
moods. A direct consequence of this is that explanations in terms of charac­
ter traits connect with emotions, which are reason-responsive states. By
contrast, explanations in terms of temperaments connect with moods, which
are a-rational. And this nicely explains why the language of character traits,
as opposed to the language of moods and temperaments, is on the whole
never used to provide excuses. Indeed, the fact that a character trait consists
in a disposition to respond affectively in a specific way to specific kinds of
370   Social Science Information Vol 48 – no 3

reasons explains why character traits often connect with praise (when we
perceive the weight the reason has for the subject as appropriate) and blame
(when we perceive it as inappropriate). In addition, the depth of the contrast
between the two types of explanations can be seen in the fact that reference
to character traits has a normative dimension that is clearly absent from
temperaments and moods.14 Thus, from the third- and first-person points of
view, character traits are understood as requiring the subject to feel and act
in such and such a way, provided no defeater is present. If he is cruel, then
this means he will apprehend the situation as giving him a reason (obviously
not a moral reason) for a certain response; if he is kind, for another. Again,
while mention of temperaments creates expectations with regard to moods
and behaviour, we do not think of them as something the subject is commit­
ted to, given the temperament. In a nutshell, since character traits are distinc­
tive dispositions to experience emotions – which have appropriateness
conditions – they are inside the space of reasons. By contrast, since tem­
peraments are dispositions to experience moods – which have no such con­
ditions – they stand outside this space.

Sentiments and emotions

What are sentiments, or what psychologists often call attitudes? We shall

now argue that sentiments are, in many important respects, very close to
character traits and have the same explanatory role. The fundamental dis­
tinction, we claim, is the following. Whereas someone with a given charac­
ter trait is affectively attached, indifferent or averse to a given value, a person
with a given sentiment is affectively attached or averse to specific things,
typically persons, but also animals, artefacts and institutions.
The contemporary use of the term ‘sentiment’ is loose. It is sometimes
employed as a synonym for the emotions or for affect in general, and is even
used in connection to beliefs that have not entirely solidified. One central and
venerable use of the term, however, refers to love and hate; and this use
broadly conceived is our concern here. Attributions of love and hate desig­
nate particularly salient forms of attachment and aversion to objects. These
are often traceable to repeated emotional interactions with these objects,
which have crystallized in distinctive long-standing affective orientations
towards them.15 Indeed such attachments and aversions come in many differ­
ent forms. The affection you may have for your dog, your devotion to your
God, your dislike of the government, and your strong taste for Byzantine
architecture all exhibit the required structure to qualify as sentiments. How
are such affective phenomena to be understood?
Deonna & Teroni Taking affective explanations to heart   371

As with character traits, reference to sentiments has the function of con­

necting in a stable and coherent fashion a cognitive and a motivational
dimension of a subject’s psychology. On the cognitive side, ascription of
sentiments goes hand in hand with a proclivity to evaluate situations in spe­
cific terms. The lover will thus apprehend situations involving the beloved
in a way not shared by those lacking his sentiment. In the most abstract
terms, she will evaluate positively situations that positively affect the
beloved, negatively situations that negatively affect him. And the situation
will be exactly the reverse if the sentiment is hate rather than love. On the
motivational side, we conceive of someone with a given sentiment as having
a specific motivational structure: the person who loves another is motivated
to further the other’s interests, to help her in case of need, to act in the light
of the other’s interests even when they conflict with his, etc.; while the hat­
ing person will be motivated to inflict damage on the other, even when this
conflicts with her own interests, etc.
Thus, like character traits, sentiments explain by ascribing a stability and
coherence to the weight subjects give in their actions and practical reasoning
to specific kinds of reasons. Moreover the relevant notion of weight can be
understood in terms similar to those given with respect to character traits.
Sentiments are also dispositions to feel given types of emotions in given types
of circumstances. For the claim that the cognitive dimension of sentiments is
affective and therefore that the mode of presentation of the relevant circum­
stances is specific to a subject with a given sentiment is at least as strong as it
is in the case of character traits. The situations confronting the beloved take
for the lover the shape of distinctive affective reactions to the circumstances
that will not be shared by the one who hates or is indifferent. The defamation
of his church will anger the devout in a way that might fill the infidel with joy.
Like character traits, sentiments exhibit a unity it is not possible to account for
via a ‘division of labour’ between beliefs and generic desires.
This is not to say that sentiments are character traits, however. They differ
in at least one fundamental respect. In character traits, the emotional profile
qualifies as an attachment to a value, whereas in sentiments it qualifies as an
attachment to an object. That is, when we refer to attachments or aversions
such as love and hate in explaining a given piece of behaviour, the explana­
tion proceeds in terms of the subject being disposed to give a distinctive
weight in his behaviour and reasoning to considerations affecting a given
object. For the kind person to give a distinctive weight to the well-being of
others is for her to be emotionally affected by this value in distinctive ways.
For the lover to give a distinctive weight to the beloved is for him to be emo­
tionally affected by this person in distinctive ways. Of course the fact that
sentiments connect primarily with dispositions to experience emotions means
372   Social Science Information Vol 48 – no 3

that sentiments, like character traits, are distinctive kinds of evaluative sensi­
tivities. In character traits, one emotionally reacts in distinctive ways to how
a given value one is attached to fares; in sentiments, one emotionally reacts
in distinctive ways to how a given object fares. While the honest person will
feel uncomfortable or perhaps outraged at the presence of injustice, the lover
will rejoice or perhaps feel pride at the success of his beloved.
Being dispositions to experience emotions, and emotions being reason
responsive, sentiments are similar to character traits in giving rise to psycho­
logical explanations squarely within the space of reasons. With sentiments,
we again face the normativity we claimed to attach to explanations in terms
of character traits. From the third- and first-person points of view, a senti­
ment is understood as a sensitivity to distinctive kinds of reasons to feel and
act. Sentiments, then, have to do with distinctive dispositions to take situa­
tions involving an object as giving reasons to feel. If Maria loves Mike, she
is disposed to apprehend given situations involving him as providing her
with reasons for certain responses. Similarly, if I hate the government, then
I am disposed to apprehend situations involving it as giving me reasons to
feel and act in distinctive ways. This is why the emotions a subject with a
given sentiment will feel are not merely expected but demanded of her. And
because these reasons, in the case of sentiments, have primarily to do with
the presence of a given object and not with a given value, the connection of
sentiments with praise, blame and excuses is more complex than in the case
of character traits. We often praise or blame people for affectively reacting
or not to reasons relevant to a given sentiment, and for doing or not doing
what we think these reasons should lead them to do; but, in contrast with
character traits, we also refer to sentiments to provide excuses, i.e. they
sometimes seem to explain but not to justify. This being said, providing an
excuse by mentioning a sentiment is not at all similar to providing one in
terms of moods and temperaments. An excuse of the former kind consists in
invoking the fact that the subject had bone fide reasons that explain why she
acted as she did but which, in the circumstances, made her oblivious to other
relevant reasons. And this sharply contrasts with excuses of the latter kind,
which do not connect at all with reasons.

Answering the remaining worries

We have given reasons to think of temperaments as providing non-rational

explanations, and of character traits and sentiments as providing reason-
sensitive explanations. Both character traits and sentiments, we have argued,
explain behaviour by reference to individual emotional profiles, and differ
Deonna & Teroni Taking affective explanations to heart   373

solely with respect to what these profiles revolve around: values for the
former and objects for the latter. Worries as to whether we can in this way
account for some traits that might be thought to be character traits, however,
will have legitimately arisen in the mind of the reader. We shall thus discuss
three types of problematic traits, which will allow us to uncover the full
potential of the present proposal.
First, do all traits allow for a neat division between being value-directed
and being object-directed? I am averse to Professor Jones and I hate intel­
lectualism. Presumably, the former is a sentiment, the latter a character trait.
But it also happens not surprisingly that I hate academics. What are we to say
about this last type of trait? Philosophers sometimes call them emotional
dispositions, other examples of which are Mike’s fear of dogs and Michelle’s
distaste for body-builders. There are many possible answers here, two of
which are the following. We might say that whether we face a sentiment or a
character trait depends on the way the subject apprehends the relevant object.
If Michelle construes body-builders as something she has a strong distaste for
because, for example, she construes them all to be like her ex-boyfriend, we
should take it that we are dealing here with a sentiment. That is to say, her
attitude focuses on body-builders in a way that appears independent of any
evaluative property they might instantiate. But body-builders might be for
her an embodiment of, say, cheap 1980s aesthetics, in which case it will
qualify as a character trait. Alternatively, one might say that all apparently
object-directed traits are in fact directed at some values these objects instanti­
ate.16 Within this option, sentiments and character traits still differ, but the
difference will be one of degree. Perhaps the most plausible way to develop
this idea is to say that, when the values concerned are such that they can be
found across various objects and situations, we will speak of character traits.
When this is not the case, we shall be more tempted to speak of sentiments.
Since both options make sense in their own way of the place of emotional
dispositions within the proposal, this should count in its favour.
Second, all-encompassing or very broad traits, like being honest, whole­
hearted, evil or deep, although they are often cited as paradigmatic exam­
ples of character traits, might sit uneasily with the proposal. Why? Because
appealing to sensitivity with respect to a given value is here less than
straightforward. To which value is the wholehearted or evil person sensi­
tive? The first and obvious thing to note is that ascriptions of those traits
to people will imply that they are disposed to evaluate in distinctive ways.
The problem is not so much the connection with values and evaluations as
that of specifying the relevant value, an apparent difficulty we think should be
addressed as follows. A deep person might prize the holy, the complex, the
sophisticated, etc.; and attachment to some of these values might correspond
374   Social Science Information Vol 48 – no 3

to more fine-grained character traits. Sensitivity to the holy might be indi­

cated by reference to her depth, but could have been pointed out by calling
her spiritual. What these examples illustrate, then, is that we sometimes
attribute ‘meta-traits’, i.e. character traits that might be realized in differ­
ent guises by more fine-grained traits and thus sensitivities to more spe­
cific values. Far from going against the present proposal, the existence of
meta-traits suggests that we think of in principle separable affective sensi­
tivities as nevertheless tending to go with one another in interesting and
informative ways.17
Third, and finally, the proposal might be thought problematic in the light
of what we may call, for lack of a better term, behavioural traits. Being
punctual, efficacious, quick, charming, distracted or available might all be
cases in point, in short, ways of being and acting that cannot be easily
thought of in terms of sensitivity to values. What is striking in this sample
of traits is that they deal exclusively with stability in observable behavioural
outcomes. They clearly explain in dispositional terms but are completely
silent as to the nature of the disposition at stake. And this is why we did not
elect them as paradigmatic cases of character traits, as we insisted that we
should agree that not all behavioural outcomes traceable to a disposition
qualify as character traits. Considering now these traits in terms of our pro­
posal, we can see their relation with character traits in an illuminating way.
Behavioural traits exhibit a disjunctive structure with respect to what
explains their obtaining in a given individual. Thus, describing someone as
punctual or quick is to focus on patterns of behavioural outcomes while
remaining neutral as regards their psychological underpinnings. I may be
punctual just because keeping my schedule does not create any problem for
me. But if my systematically being on time is due to an affective attachment
to, say, common courtesy, then my punctuality should be explained in terms
of my being a courteous person – a character trait. This means that the pro­
posal supplies a very valuable distinction between stable behavioural out­
comes on the one hand and character traits on the other. We have more than
once insisted on the fact that the former can be present without the latter, and
of course the reverse is also, to our regret, common. For, as already noted,
character traits do not operate in isolation: courteous people can also be lazy
or obedient, other character traits that on occasion will trump courtesy. As
we have been at pains to argue, although character traits do manifest them­
selves in typical behavioural outcomes (courtesy by courteous behaviour),
very often the best we do in manifesting the trait is to show regret or shame
that we have not been courteous, or express our hope that we will be courte­
ous on the next occasion. The upshot is that, while sceptics regarding char­
acter traits are right to point out how unreliable character traits are in
Deonna & Teroni Taking affective explanations to heart   375

predicting their typical behavioural outcomes – and the evidence adduced on

that count is indeed impressive – they are wrong in conceiving this as show­
ing that we lack character traits. What we lack is rather character, in the
sense that we are pretty bad at living up to the values we are attached to. In
short, behavioural traits do not make up a self-standing psychological cate­
gory, although they might very well be used, and in context be taken, to refer
to character traits.18
Julien Deonna is working on the theories of emotions, moral psychology and philosophy of
psychology. He has published in all these areas. He is Assistant Professor at the University
of Geneva and Scientific Collaborator with the National Centre for Research in the
Affective Sciences. Author’s address: University of Geneva, 2 rue de Candolle, 1205
Geneva, Switzerland. [email: Julien.Deonna@lettres.unige.ch]

Fabrice Teroni is working in the philosophy of mind, more specifically on memory and the
emotions, and in epistemology. He has published both on the theory of emotions and on
memory. He is currently Lecturer at the Universities of Geneva and Bern, and is Associated
Researcher within the National Centre for Research in the Affective Sciences. Author’s
address: University of Geneva, 2 rue de Candolle, 1205 Geneva, Switzerland. [email:
Fabrice.Teroni@lettres.unige.ch]

Notes
We would like to thank F. Lauria, K. Mulligan, A. Pé-Curto and G. Peebles, as well as an
anonymous referee, for their comments and criticisms on previous versions of this article.
We also acknowledge the support of the Swiss PNR in Affective Sciences in the writing of
this article.

  1. For an empirical approach to character traits that warrants scepticism about their rele­
vance in psychological explanation and prediction, see e.g. Ross & Nisbett (1991). For the
lessons drawn from this scepticism in philosophy and ethics, see Doris (2002) and Harman
(2000).
  2. The further difficult question as to whether the relevant episodes enjoy explanatory prior­
ity over the dispositions, or vice versa, will have to be addressed at another occasion.
  3. While not universally accepted, the conception developed here is part of a well-estab­
lished way of conceiving of the emotions as forms of evaluation (see in particular Tappolet,
2000; Scherer, 2001; Roberts, 2003; Deonna, 2006) and therefore as subject to rational con­
straints (e.g. De Sousa, 1987; Mulligan, 1998; Teroni, 2007).
  4. It is not rare to meet the view that moods are apprehensions of ‘the whole world’ or some
vaguely specified aspects of it (Goldie, 2000: 141) in some evaluative way. While not deprived
of any intuitive plausibility – it might be one way of affirming that moods colour our experience
– we believe that taken at face value this view should be given up since it implies that moods
should almost always be assessed as inappropriate, rather than being, as is so much more plau­
sible, not subject to this form of normative evaluation.
  5. Of course, this does not imply that moods cannot constitute reasons when they feature in
the content of other mental states. For instance, a prolonged bad mood might worry me to the
point of going to the doctor. The mood is here the object of my worry and is evaluated by it as
376   Social Science Information Vol 48 – no 3
a threat to my health. It is only by virtue of being part of the content of the emotion that the
mood gives me a reason to go to the doctor.
  6. For a subtle discussion of the various interactions between moods and emotions, see
Goldie (2000: 143–51).
  7. Although we do not distinguish character from personality traits, the following should be
kept in mind. On the whole, philosophers and psychologists aim to capture the same phenom­
enon by the use of these two terms (e.g. Doris, 2002: 19). Both connect with stability and
consistency in behavioural outputs, but, as our own account will bear witness, the philosophers’
notion tends to refer to a subclass of these stabilities and consistencies, those grounded in the
subject’s evaluative outlook. Terminology notwithstanding, we tend to side with Goldie (2004)
in regarding character traits as denoting deeper facts about people’s psychology than personal­
ity traits.
  8. For an important recent endeavour along these lines, see Zagzebski (1996).
  9. The two dimensions of character traits constitute a crucial aspect of Hudson’s analysis
(1980). The claim that they are independent from one another, a claim we take issue with, is
salient in Butler’s discussion (1988).
10. Essentially conative accounts of character traits can be found in Brandt (1988) and
Harman (1999).
11. For a detailed description of the mechanics involved in patterns of emotional disposi­
tions, see Helm (2001).
12. The idea that subjects with a given character trait lend a specific weight to distinct con­
siderations, reasons or values is an important theme in the writings of Goldie (2004), McDowell
(1979), Morton (1980) and Wiggins (1975–6).
13. The reader interested in the kind of issues alluded to here in connection with the masking
problem for dispositions should consult Johnston’s (1992) seminal contribution and the discus­
sions it has given rise to.
14. This normative dimension of character traits is especially salient in Goldie’s discussions
(2000, 2004). Be aware, though, that his use of the term ‘excuses’ is broader than ours.
15. See Broad (1954–5: 297–8) for a subtle description of the genesis of sentiments.
16. This connects with a very old and ongoing debate as to whether love and other sentiments
are attachments to objects independently of the values we perceive in them. For a good survey
of various treatments of this issue, see Soble (1989).
17. And this in turn raises the interesting traditional issue, usually discussed in connection
with the virtues, about the extent to which various character traits form or should form broader
units.
18. For further criticisms of the sceptical literature on character traits, some of which go
along the same lines as those alluded to here, see in particular Goldie (2000, 2004), Kamtekar
(2004), Merritt (2000) and Sreenivasan (2002).

References
Brandt, R. (1988) ‘The structure of virtue’, Midwest studies in philosophy 13: 64–82.
Broad, C. D. (1954–55) ‘Emotion and sentiment’, in D. Cheeney (ed.) Broad’s critical essays
in moral philosophy, pp. 283–301. New York: George Allen & Unwin.
Butler, D. (1988) ‘Character traits in explanation’, Philosophy and phenomenological research
49(2): 215–38.
Deonna & Teroni Taking affective explanations to heart   377
De Sousa, R. (1987) The rationality of emotion. Cambridge, MA: MIT Press.
Deonna, J. (2006) ‘Emotion, perception and perspective’, Dialectica 60(1): 29–46.
Doris, J. (2002) Lack of character. New York: Cambridge University Press.
Goldie, P. (2000) The emotions. Oxford: Oxford University Press.
Goldie, P. (2004) On personality. London: Routledge.
Harman, G. (1999) ‘Moral theory meets social psychology: virtue ethics and the fundamental
attribution error’, Proceedings of the Aristotelian society 99: 315–31.
Harman, G. (2000) ‘The nonexistence of character traits’, Proceedings of the Aristotelian
society 100: 223–6.
Helm, B. (2001) Emotional reason. Cambridge: Cambridge University Press.
Hudson, S. (1980) ‘Character traits and desires’, Ethics 90(4): 539–49.
Johnston, M. (1992) ‘How to speak of the colors’, Philosophical studies 68: 221–63.
Kamtekar, R. (2004) ‘Situationism and virtue ethics on the content of our character’, Ethics
114: 458–91.
McDowell, J. (1979) ‘Virtue and reason’, The monist 62(3): 331–50.
Merritt, M. (2000) ‘Virtue ethics and situationist personality psychology’, Ethical theory and
moral practice 3: 365–80.
Morton, A. (1980) Frames of mind. Oxford: Oxford University Press.
Mulligan, K. (1998) ‘From appropriate emotions to values’, The monist 84(1): 161–88.
Roberts, R. (2003) Emotions. Cambridge: Cambridge University Press.
Ross, L. & Nisbett, R. (1991) The person and the situation: perspectives of social psychology.
New York: McGraw-Hill.
Ryle, G. (1949) The concept of mind. London: Penguin.
Scherer, K. (2001) ‘Appraisal considered as a process of multilevel sequential checking’, in
K. Scherer, A. Schorr & T. Johnson (eds) Appraisal processes in emotion theory, methods,
research, pp. 92–120. Oxford: Oxford University Press.
Soble, A., ed. (1989) Eros, agape, and philia. St Paul, MN: Paragon House.
Sreenivasan, G. (2002) ‘Errors about errors: virtue theory and trait attribution’, Mind 111:
47–68.
Tappolet, C. (2000) Emotions et valeurs. Paris: Presses Universitaires de France.
Teroni, F. (2007) ‘Emotions and formal objects’, Dialectica 61(3): 395–415.
Wiggins, D. (1975–6) ‘Deliberation and practical reason’, Proceedings of the Aristotelian
society 76: 43–9.
Zagzebski, L. (1996) Virtues of the mind. Cambridge: Cambridge University Press.