How should neuroscience study emotions? by
distinguishing emotion states, concepts, and
experiences
Ralph Adolphs
Division of Humanities and Social Sciences, California Institute of Technology, HSS 228-77, Caltech, Pasadena,
CA 91125, USA. E-mail: radolphs@caltech.edu
Abstract
In this debate with Lisa Feldman Barrett, I defend a view of emotions as biological functional states. Affective neuroscience
studies emotions in this sense, but it also studies the conscious experience of emotion (‘feelings’), our ability to attribute
emotions to others and to animals (‘attribution’, ‘anthropomorphizing’), our ability to think and talk about emotion
(‘concepts of emotion’, ‘semantic knowledge of emotion’) and the behaviors caused by an emotion (‘expression of
emotions’, ‘emotional reactions’). I think that the most pressing challenge facing affective neuroscience is the need to
carefully distinguish between these distinct aspects of ‘emotion’. I view emotion states as evolved functional states that
regulate complex behavior, in both people and animals, in response to challenges that instantiate recurrent environmental
themes. These functional states, in turn, can also cause conscious experiences (feelings), and their effects and our
memories for those effects also contribute to our semantic knowledge of emotions (concepts). Cross-species studies, dissociations
in neurological and psychiatric patients, and more ecologically valid neuroimaging designs should be used to
partly separate these different phenomena.
Key words: emotion; concepts; affective neuroscience; amygdala; feelings
The neuroscientific investigation of emotion, affective neuroscience,
is one of the most interesting and vibrant new disciplines,
and of paramount importance for understanding individual differences
and many psychiatric disorders. It is also one of the
most confusing disciplines, in large part because the word
‘emotion’ is used in multiple ways. In this debate with Lisa
Feldman Barrett, we will try to make it less confusing. I begin
with a bare-bones description of my view (cf. Figure 1), then discuss
how confusions arise, and conclude with some thoughts
on how the neuroscientific investigation of emotions could
move forward. Since my focus is on how even to think about
‘emotion’ as scientists, this brief article is light on reviews of
empirical studies and heavier on conceptual material. I aim
here to present my own view as clearly as possible, and will reserve
most of my comments on Lisa’s view for the rebuttal.
Affective neuroscience, and cognitive neuroscience more
generally, requires a close interplay between the vocabularies
and frameworks of two different scientific disciplines: psychology
and neuroscience. When they study emotions, these two
disciplines are trying to describe the same states or processes (I
use these terms interchangeably here), but often on the basis of
different kinds of data, methods, and theories. Some views suppose
one discipline has priority over the other, often that neuroscience
trumps psychology: if we can’t find evidence for a
psychological construct from brain data, then the psychologists
were wrong about that construct. I want to resist such a view, in
large part because we really have very little idea about how to
interpret neuroscience data, so whatever evidence it does or
does not provide for a psychological theory should be considered
extremely preliminary. So while I believe that emotions
Received: 9 July 2016; Revised: 9 July 2016; Accepted: 11 October 2016
VC The Author (2016). Published by Oxford University Press.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/
licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
For commercial re-use, please contact journals.permissions@oup.com
24
Social Cognitive and Affective Neuroscience, 2017, 24–31
doi: 10.1093/scan/nsw153
Advance Access Publication Date: 19 October 2016
Original article
Downloadedfromhttps://academic.oup.com/scan/article-abstract/12/1/24/2624554byMasarykovaUniverzitauseron27February2020
are brain states, I also believe that we need to begin by understanding
them as psychological states.
What are emotions? Emotions are functional
states, implemented in the activity of neural
systems, that regulate complex behaviors
My view of emotions begins with everyday observations and
concepts (but it doesn’t end there). Clear instances of an emotion
state are those that Charles Darwin already noted in his
book, The Expression of the Emotions in Man and Animals
(Darwin, 1872/1965). That is, at least in the first instance, we attribute
emotion states to people and to many other animals on
the basis of their context-dependent observed behavior.
Emotion states, together with many other mental state attributions,
provide causal explanations of behavior. I take it that
there is little disagreement that a weeping adult, a screaming
child, and a hissing cat all are in various emotion states. What
exactly we want to call those states, and whether they should
be thought of categorically or dimensionally, are further and
more complicated questions. For simplicity of exposition I will
use words like ‘fear’ to refer to the hissing cat in this article,
e.g.—but the meaning of that term will likely need to be revised
to be useful in a mature affective neuroscience. It is also the
case that the sets of stimuli and behaviors that specify an emotion’s
functional role are highly context-dependent, and no single
behavior is typically diagnostic. For instance, the cat’s
hissing behavior by itself is also quite compatible with an emotion
state of anger, rather than fear (or both)—one would need
to do further experiments (challenging the animal with specific
stimuli, observing its behavior) to disambiguate this, since the
hissing is not unique to fear. Likewise, the weeping adult may
be weeping from sadness, or anger, or some combination of
these emotion states (or indeed just be acting)—we would need
more information about their facial expression, their other behavior,
and the circumstances in which the behavior is
observed. This applies to most emotional behaviors: they help
narrow down the set of possible emotion states that explain the
behavior, but we typically require considerable additional
information about the history and circumstances under which
the behavior is exhibited to disambiguate between multiple
possible emotions states that could account for the behavior.
We do this all the time: if somebody is behaving in a way indicative
of an emotion state, we usually probe them further to get
additional evidence (for instance, by asking them how they feel,
or what happened, or what they intend to do). The contextdependency
of emotion states is also critical to consider for
affective neuroscience studies in which we want to experimentally
manipulate emotion states.
If the above simple starting point is a reasonable place to
begin to develop a scientific concept of emotion states, the next
question is what it is about these examples—the weeping person,
screaming child or hissing cat—that distinguish them as
evidence for emotion states. Many behaviors are caused by central
states of various kinds: so what distinguishes emotion
states? A useful comparison is with behaviors that are either
simpler or more complex. Reflexes are simpler than emotional
behaviors. Reflexes are relatively rigid and typically do not
interface in a rich way with other psychological states—they do
not need to interact with attention or memory, for instance.
They just connect sensory inputs to motor outputs (the reality
is more complicated, but let’s simplify for the sake of the examples).
So emotions are more complex than reflexes, they ‘decouple’
stimuli from responses, thus affording much more
flexibility (Scherer, 1994). Planned, volitional behavior, on the
other hand, is more flexible and more complex than emotions.
Emotional behaviors are not like that either—they don’t have
that many degrees of freedom. Emotions regulate behavior at a
level of complexity intermediate to that of reflexes and volitional
behavior (Adolphs, in press). Charles Darwin had a similar
notion in mind when he wrote about emotional behaviors,
‘Nor can these movements in the dog be explained by acts of
volition or necessary instincts, any more than the beaming eyes
and smiling cheeks of a man when he meets an old friend.’
(Darwin, 1872/1965)
In my view, then, emotion states evolved in order to allow
us to cope with environmental challenges in a way that is more
flexible, predictive and context-sensitive than are reflexes, but
that doesn’t yet require the full flexibility of volitional, planned
behavior. They evolved to deal with particular, recurring themes
in our environment; and because most of the specific sensory
features of those themes are highly variable, they also critically
involve learning. Broadly, emotions are one solution to determining
what is relevant in the world by learning recurring patterns—themes,
if you will. In fact, I think that the patterns to
which emotions are tuned are at the level of ‘core relational
themes’ (Lazarus, 1991), even if the specific relational themes
that psychological theories have proposed so far may not be the
right ones. This is a functional definition of emotion states
(I use the term ‘functional’ in the philosophical sense of functionalism,
not in the developmental psychological sense). A
functionally defined term is defined by what it does rather than
by how it is constituted. Consequently animals with very different
brains—and, in principle, even robots—could nonetheless
be in similar emotion states.
Building on the comments above, we can begin to list the
properties that emotion states exhibit, which psychological theories
of emotion have often attempted to do, and which my colleague
David Anderson and I have also attempted (Anderson
and Adolphs, 2014). Figure 2 summarizes some of these in a provisional
list. One universally recognized feature of emotions is
that they can be related to one another in a similarity space.
The simplest such space has two dimensions (perhaps
Not at all Very much so
Modular.
Localized.
Distributed.
Cortical.
Subcortical.
Innate.
Acquired.
Domain-general.
Domain-specific.
Fig. 1. Rate your position. Which applies to emotions? Indicated are my own (blue),
my take on Lisa Barrett’s (Barrett, 2006; Wilson-Mendenhall et al., 2011) (pink) and
my take on Jaak Panksepp’s (1998) (gray), to provide three different views (any errors
are of course mine). Many of the terms have unclear meanings, and the figure is intended
only to give a rough starting point for discussions, not to quantify theoretical
frameworks. Lisa saw a prior version of this figure and sent some corrections to my
original take on her view. My original depictions of her positions are indicated by
circles; the corrected positions from Lisa are denoted by triangles.
R. Adolphs | 25
Downloadedfromhttps://academic.oup.com/scan/article-abstract/12/1/24/2624554byMasarykovaUniverzitauseron27February2020
corresponding to something like valence and arousal), although
we will need additional dimensions to capture all the varieties
of emotions. Fairly discrete clusters of emotion instances in this
space would then correspond to specific emotion categories.
Another prominent feature of emotion states is their flexibility,
which derives from their ‘decoupled’ nature: they are central
states that persist for some time, and so can accumulate a host
of contextual information before triggering behavioral decisions.
The persistence of an emotion state also permits it to
interface in a rich way with the rest of cognition, a major topic
of study (Pessoa , 2013; Okon-Singer et al., 2015). Given the multiple
causal effects of an emotion state, these need to be coordinated
in some way, another feature that is frequently
emphasized in psychological theories of emotion. Finally, emotion
states have pre-potent control over behavior, a feature
similar to historical schemes of an ‘interrupt’ mechanism that
can terminate ongoing goal-directed behavior when a sudden
environmental challenge is encountered (Simon, 1967). All of
this is of course further complicated in adult humans, since
there is some degree of volitional control over emotion states
and their expressions. Emotion regulation and strategic/deceptive
signaling through emotional expression are perhaps the
most distinctively human aspects of emotion.
The features of emotion states sketched in Figure 2 are relatively
domain-general, but in their combination provide clues to
the domain-specific roles that guided the evolution of particular
emotions. The way that many of the features are engaged relative
to specific stimuli and behaviors will demarcate emotion
categories. For instance, conditioned taste aversion or Pavlovian
fear conditioning both involve learning (by itself a domaingeneral
feature), but they apply to specific kinds of stimuli—not
just any stimulus can be conditioned in this way, and not just
any behavior can be produced (only those stimuli, and behaviors,
relevant to dealing with threats and to avoiding poisonous
foods, in this example). In many cases, the possible functional
roles that an adult human emotion can play are enormous, so I
believe that we should begin the investigation by hypothesizing
the core functional roles that specify the emotion category. This
is one reason that developmental and comparative data are essential.
They can give us some hints as to what the functional
relations are that guided the evolution of the emotion—this is
the ultimate functional story we would want to know: what
functions, in our ancestral environment, did an emotion play
that resulted in the selection of neural mechanisms to implement
that function? Evolutionary psychology tries to provide
precisely such functional accounts of emotion states (Tooby
and Cosmides, 2008), including not only accounts for emotions
like fear and disgust, but also for social emotions: these serve
functional roles in regulating our social behavior (Sznycer et al.,
2016). One exciting approach in affective neuroscience could be
to design experiments that engage functional roles for specific
emotions as hypothesized by theories from evolutionary
psychology.
I think many of the attributes of emotion states, and the
functional conception of them that I am advocating, would put
me in the ‘Basic Emotion’ camp, even though I would disagree
with many other details of certain Basic Emotion theories. By
‘basic’ I mean ‘biologically basic’ (Ortony and Turner, 1990), i.e.,
a category defined in an evolutionary sense. I take it that this is
also the sense in which psychologists like Ekman (1999) have
used the term, and the sense in which neurobiologists like
Panksepp (1998) have used the term (although they posit different
sets of basic emotions). Basic Emotion theorists typically
have not only a list of the specific features an emotion must
Scalability: An emotion can scale in intensity, often conceived
of as an arousal dimension in many psychological theories.
Parametric increases in the intensity of an emotion can recruit
discrete behaviors, such as the transition from hiding to fleeing
during the approach of a predator.
Valence
of behavior (whether to behave at all, and how vigorously),
valence describes a second fundamental aspect: whether to
approach or to withdraw.
Persistence: An emotion state outlasts its eliciting stimulus,
unlike reflexes, and so can influence cognition and behavior
(and other emotions) for some time.
Learning: While some stimuli innately induce emotions, the
emotional significance of the vast majority of stimuli needs to be
learned. Pavlovian fear conditioning and conditioned taste
aversion would be two examples (for fear and disgust,
respectively) that are well studied in the lab.
Priority over behavioral control: Emotions are prepotent in
their control over behavior, requiring additional regulatory
mechanisms to override their behavioral expression.
Poised for social communication: Emotional behaviors in
most animals are typically honest signals of states that predict
behaviors. As such, they have been co-opted as signals of
which others can take advantage (conspecifics as well as
predators and prey).
Hierarchical behavioral control: As Tinbergen first observed,
many behaviors come as packages that can be controlled
hierarchically. Emotions implement their effects in this way.
Multi-component effects: Most psychological theories
emphasize that emotional behaviors consist of multiple
components, all of which need to be coordinated. Emotion
states accomplish this coordination.
Similarity structure
Flexibility
Coordination
Automaticity
Fig. 2. Features of an emotion. These features underpin the functional properties of emotions, and may help define dimensions or categories of emotions in a mature
theory of emotion; they are not necessary and sufficient conditions, and the list is no doubt incomplete. Several of these are modified from (Anderson and Adolphs,
2014).
26 | Social Cognitive and Affective Neuroscience, 2017, Vol. 12, No. 1
Downloadedfromhttps://academic.oup.com/scan/article-abstract/12/1/24/2624554byMasarykovaUniverzitauseron27February2020
satisfy in order to count as basic (e.g. culturally universal, specific
physiological profile etc.) but also supply their inventory of
the particular basic emotions (e.g. happiness, fear, anger etc.).
Although I think that we can indeed begin to list features (cf.
Figure 2), I am reluctant to actually name a list of emotions, because
I do not think we have enough data yet to do this (especially
cross-species data), and because the words for emotions
that already exist are likely to be misleading. I also see no reason
why a specific emotion category could not also be represented
in a dimensional space (discrete and dimensional ways
of describing emotions seem complementary).
There is a lot of work to be done in order to figure out the
functional role of different emotions, and in order to come up
with the best categories and/or dimensions by which to characterize
the different emotion states. But I think we also have a lot
of data already that points the way for how best to parse emotions.
Some of those data are from psychological studies in
humans, some from behavioral studies in animals, and some
from neuroscience studies in both species. Perhaps the best
place to start, if one wanted to begin neuroscientific studies of a
specific emotion, would be to pick an emotion on which there is
a fair amount of evidence across all these different sources.
Emotion categories like anger, fear or disgust seem particularly
well suited, for instance.
How to get confused about emotions
Affective neuroscience can be confusing when it fails to make
distinctions between different aspects of affective processing.
The titles of papers and discussions that authors give are often
no help here either, since they frequently conflate different
meanings of the term ‘emotion’. The most common ambiguity
is between ‘emotion’ as conceived above (the functional state)
and its conscious experience, conceptualization or attribution.
Generally, when I am in a state of anger, I also feel angry, and I
also think about being angry. Those are all interesting processes
to study, but they are distinct. I am interested here, in the first
instance, in how we should study ‘emotion states’, not ‘how
people can use concepts to think about emotions’ or ‘how people
make attributions of emotions’, or ‘how people can speak
about emotions’. Those are all further interesting questions,
and certainly questions that affective neuroscience should investigate,
but I don’t believe they are the place to start because
they don’t help us to ground what emotions are supposed to be
about in the first place.
Take again the example of my hissing cat. The cat cannot
speak about emotions, plausibly also cannot think about nor
has a concept of what an emotion is, and it remains unclear
how to determine if it would even have a conscious experience
of an emotion, whatever one means by that exactly. But it is
clear to me that it has emotions (in the above functionally
defined sense)—this is what I attribute to the animal in order to
explain and predict its behavior, and indeed it works fairly well
most of the time. The emotion states are the internal functional
states that produce the behaviors we see. Colloquially, that is
what we would say about the cat: it is in a ‘state of fear’.
Affective neuroscience investigations of these states would
then study what it is in the brain of the hissing cat that causes
those behaviors, in response to particular context-dependent
stimuli. Of course, there are many more subtle and mixed emotions
than these blatant examples, as well as more sustained
states that we would typically call moods, and there can be further
debate about where to draw the line and say that we would
no longer call a state an emotion state. To anchor the
investigation, however, I believe we need clear, strong emotion
states evoked by ecologically valid stimuli, where ‘ecologically
valid’ simply means ‘experimentally re-creating the functional
challenge that we hypothesize engages the emotion under
investigation’.
Emotion states, then, are not the same as emotion concepts
or emotion experiences. By analogy, if I wanted to study people’s
concepts of planets, I could do psychological or neuroimaging
studies to investigate this (I would study what people
know and think about planets). But if I want to study planets, I
would do astronomy and use a telescope. Just like concepts of
planets are not planets, concepts of emotions are not emotions.
Emotion states are also not the same as conscious experiences
of emotions. In this respect, the usage of the word ‘emotion’
that I am advocating is similar to how we use words like ‘vision’
or ‘memory’ in neuroscience. For the layperson, vision and
memory are all about conscious experiences (seeing and remembering).
For the scientist they are functionally defined
terms, and indeed we now know that both visual processing
and memory can be non-conscious (as in blindsight and nondeclarative
memory). Our commonsense concepts for most
mental-state terms seem to depend on our concept of conscious
experience, but I think our scientific concepts for mental-state
terms should not. If they did, it becomes problematic how to
study the minds of people and animals who cannot use language
to tell us about their conscious experiences. Joseph
LeDoux has correctly pointed out this problem: if we use the
commonsense concept of ‘fear’ when describing animal neuroscience,
we risk confusing this with the attribution of a conscious
experience of fear. LeDoux concludes from this that we
should stop using words like ‘fear’ or ‘emotion’ when doing animal
neuroscience (LeDoux, 2012), but I think there is a simpler
solution: do the same thing we as scientists do when we study
vision or memory. Use a scientific concept of ‘emotion states’
that is not based on conscious experience.
To summarize how people get confused about what is meant
by ‘emotion’: there are distinctions between the functional
emotion state (‘the emotion state’), its conscious experience
(‘the experience of the emotion’), our ability to attribute emotions
to others and to animals (‘attribution of emotion’; ‘emotion
perception’), our ability to think and talk about emotion
(‘conceptualizing emotion’), and the behaviors caused by an
emotion state (‘the expression of emotions’, ‘emotional reactions’)
(Table 1). I think emotions are first and foremost about
the first of these, and all the others are derivative (but no less
interesting to study).
Dissociating emotion states from emotion
concepts: an example from neuroscience
An example of how emotion states can be separated from conceptual
knowledge of emotions comes from classical cognitive
neuropsychology, the use of neuroscience data in order to help
make distinctions by showing dissociations (Caramazza, 1986).
The famous patient S.M. (Feinstein et al., 2016), who has bilateral
lesions to her amygdala, shows a dissociation with respect to
fear that is about as good as dissociations get. She can laugh,
she can cry and she can endorse feelings of happiness and sadness
and most other emotions. But she does not show any of
the effects of a state of fear that we would normally use in order
to attribute fear (Feinstein et al., 2011). She does not show normal
avoidant behaviors to threatening situations, she does not
show autonomic responses or give subjective ratings of fear to
R. Adolphs | 27
Downloadedfromhttps://academic.oup.com/scan/article-abstract/12/1/24/2624554byMasarykovaUniverzitauseron27February2020
normally fear-inducing stimuli, and she fails to show learning
based on unconditioned fear in Pavlovian fear conditioning. A
subset of the same deficits (minus the subjective ratings and
with simpler behaviors) is seen in animals with bilateral amygdala
lesions (Amaral and Adolphs, 2016).
Despite the complete failure to induce a state of fear from
any external stimuli, S.M. can tell us a lot about fear. She has
read books and watched movies in which fear occurs, she has
spoken with other people about fear, and she even has autobiographical
memories of feeling afraid as a child (plausibly before
her amygdala was fully lesioned) (Feinstein et al., 2011, 2016).
Consequently, she has accumulated an impressive store of semantic
knowledge about fear, so much so that we should say
that she has a concept of fear. She can tell you that people who
are afraid scream and run away; she can tell you that being
chased by a bear would make you afraid; she can use the word
‘fear’ appropriately in conversation. But she herself does not instantiate
the state of fear, even though she has so much conceptual
knowledge about fear. Just having the concept of fear is
typically insufficient for the state of fear. In fact, it is extraordinarily
difficult to induce an emotion state by just activating the
concept of an emotion. If it were easy, depressed people would
just need to think about being happy and they would be happy.
In humans, the routes by which a state of fear can be
induced are of course considerably more varied than in other
animals, and include memories and imaginings in addition to
actual occurring sensory stimuli. Indeed, if I think hard about
situations in which I would be afraid, I feel a little bit afraid. So
conceptual representations of emotions do have some effect at
least on the conscious experience of emotions, and presumably
on the emotion state as well. Conversely, being in an emotion
state typically also causes conceptual representations of emotions.
If you are in a state of fear, you typically also think about
fear and believe you are in a state of fear. So another important
challenge for affective neuroscience is to detail the causal interactions
between emotion states, emotion experiences, and
emotion concepts: in healthy adult humans, all three usually
occur together.
There is a final important dissociation that patient S.M.
showed us. As with the case of memory, where H.M. showed us
how declarative and nondeclarative memory can be dissociated,
S.M. has also shown us that fear to external stimuli, and panic
to interoceptive stimuli, can be dissociated. S.M. does panic if
she feels like she is suffocating [elicited in the experiment
through inhaling carbon dioxide (Feinstein et al. 2013)]. This was
a very interesting advance, and showed us that our scientific
concept of fear needs finer distinctions, just like H.M. and many
studies since then have given us a more fine-grained taxonomy
of memory. So the case of S.M. shows us three different
dissociations relevant to affective neuroscience: that the state
of fear can be dissociated from other emotion states; that the
state of fear can be dissociated from its concept; and that there
are varieties of fear, to which we may want to give separate
names [anxiety, fear and panic have all been used already, and
there may be additional varieties that function with respect to
specific types of threat (Silva et al., 2013)].
A framework for neuroimaging studies of
emotion states: systems, hierarchy and
topography
Explaining how an emotion state is implemented in the brain
requires us to explain which structures, at which point in time,
implement particular computations. That makes it nonsensical
to ask if ‘fear is in the amygdala’, e.g. since the state is distributed
in both space and time. Nonetheless, we can say that the
amygdala is one component of the neural system for fear, and
moreover a necessary component. At the coarsest level, we
know there’s something happening in somebody’s brain for the
several seconds or minutes they are in a state of fear. At the
most microscopic level, we know there are causal events, each
at the millisecond scale, across billions of neurons. The first description
is too low dimensional; the second is too high dimensional.
So the challenge to the neuroscientist is: can you find
something useful in the middle, something at the level of neural
systems that eventually allows us to understand how emotions
link stimuli to behavior (and other cognitive states).
Figure 2 can motivate initial hypotheses here. As already
noted, one prominent and universally acknowledged feature of
emotions is that they have a similarity structure. Anger and fear
are more similar than anger and happiness, for example.
Similarity relationships are often partly captured in a 2D space
of valence and arousal. These facts motivate the hypothesis
that there also should be similarity relationships in the brain,
and perhaps topography in how this is represented. Indeed,
studies in rodents have argued for a topography in the nucleus
accumbens that maps the dimension of valence (Berridge and
Kringelbach, 2013), at least with respect to feelings. One challenge
in discovering topography is that we will need a better description
of the dimensional space that defines similarity
relations among emotion states. On the other hand, semantic
knowledge for emotions has indeed been mapped, at least
across cortex, and can be compared with semantic knowledge
of many other concepts for which we have words (Huth et al.,
2016). Another study (Skerry and Saxe, 2015) was notable for
comparing similarity amongst neural representations with conceptual
similarity in how people rate emotions according to several
popular psychological theories (arousal/valence; basic
Table 1. Some examples of different aspects of emotion investigated in affective neuroscience, and my opinion about how central they are to a
functionally defined emotion state
Aspects of processing that are central to an emotion state.
• emotion-cognition interactions (Pessoa, 2013; Okon-Singer et al., 2015)
• emotional learning and memory (Phelps, 2006)
• eliciting strong emotions with ecological stimuli (Mobbs et al., 2010; Feinstein et al., 2013)
Aspects of processing that are less central to an emotion state.
• perceiving emotional social signals (emotion perception) (Adolphs et al., 1994)
• inferring emotions in other people (social inference, theory of mind) (Spunt and Lieberman, 2012; Skerry and Saxe, 2015]
• semantic processing about emotions (concepts) (Huth et al., (2016)
• lexical processing about emotions (words) (Rapcsak et al., 1989)
28 | Social Cognitive and Affective Neuroscience, 2017, Vol. 12, No. 1
Downloadedfromhttps://academic.oup.com/scan/article-abstract/12/1/24/2624554byMasarykovaUniverzitauseron27February2020
emotions; appraisal dimensions). That study (Skerry and Saxe,
2015) found that attributions about other people’s emotions
that we make from thinking about the situations in which people
find themselves activate representations in a system of
brain structures known to be involved in mental state attribution
more generally (such as the temporoparietal junction, dorsomedial
prefrontal cortex and precuneus); the best match
between neural and psychological similarity structures held for
ratings on appraisal dimensions. These recent examples use
techniques that would likely be useful also to investigate how
emotion states (functionally defined) might be topographically
represented in the brain [voxel-based modeling (Huth et al.,
2016), and representational dissimilarity analysis (Skerry and
Saxe, 2015), respectively).
Another reasonable hypothesis derives from the hierarchical
coordination that emotion states achieve. In this respect, they
feature what Tinbergen already observed in the fixed action
patterns many animals exhibited (Tinbergen, 1950). Some
neurobiological studies, especially in rodents, have produced
very detailed knowledge of certain components of an emotion
state; for instance, optogenetic activation of specific neuronal
populations in the hypothalamus can produce directed aggressive
behavior in mice (Lin et al., 2011). The next question is:
what are the inputs to these hypothalamic neurons that would
normally orchestrate this behavioral component? The coordination
amongst many components requires a hierarchical control
of sorts, and we could test whether this is accomplished by
yet a separate component, or whether it arises from network
dynamics amongst all the pieces. This is basically what systems
neurobiology is already doing: studying specific components of
an emotion, and trying to figure out how they are connected to
produce all the features of an emotion. Such systems-level
studies of emotions in animals try to choose ecologically valid
situations to induce emotion states like fear, or try to dissect
specific components of such states, like fear learning in
Pavlovian fear conditioning.
So we have some promising examples of topography in
human fMRI studies that were not about actual emotion states
but instead about concepts; and from the components of actual
emotion states but studied in animals. In contrast, there have
been fewer neuroimaging studies in humans that have attempted
to actually induce real emotions in human subjects,
and fewer still that have attempted to dissociate them from experiences
or conceptual processing of emotions. Nonetheless,
there have been a handful of important imaging studies, ranging
from early ones with PET (Damasio et al., 2000) to later ones
with fMRI (Mobbs et al., 2007, 2010), that derive conclusions
about particular brain structures involved in processing emotions
from trying to induce actual specific emotion states
(through autobiographical recall of emotional events, fear of
electric shock, or with innately emotion-inducing stimuli like
tarantulas, respectively). Those studies have focused on brain
structures mostly distinct from the cortical regions emphasized
in studies of semantic representation or mental state attribution.
They have instead emphasized subcortical structures like
the amygdala, hypothalamus, and periaqueductal gray, much
as have the studies of emotion states in animals (as well as additional
cortical regions such as ventromedial prefrontal cortex
and insula). All of these regions have also been noted in human
and animal lesion studies, and have been the material for several
influential neurobiological theories of emotion (Damasio,
2003; Craig, 2008).
It is worth noting that our knowledge from fMRI studies of
the neural regions encoding information about aspects of emotions
has only emerged in the last few years. Initial metaanalyses
(Lindquist et al., 2012) searching for ‘basic’ emotion
representations had little success, no doubt in good part because
the studies used for the meta-analysis were underpowered,
univariate and mixed many different aspects of ‘emotion’.
That has changed with general increase in sensitivity of fMRI
studies and the widespread adoption of multivariate analyses,
developments that produced evidence for dimensional components
like valence (Chikazoe et al., 2014), and, more recently, indeed
found evidence for some basic emotion representations
(Saarim€aki et al., 2015), conclusions replicated also in metaanalyses
of studies on emotion categories (Wager et al., 2015).
The rarity of human studies that have actually produced strong
emotion states, and the frequent conflation of emotion states,
concepts, and feelings, all suggest caution in concluding much
yet about emotion states on the basis of neuroimaging data. On
the other hand, there are complementary data from animal
studies, and there are lesion dissociations in humans, both of
which help to motivate strong hypotheses about where to look
in the brain in future neuroimaging studies of emotion states.
So to what extent can we in fact dissociate the functional
emotion state from emotional experience, labeling, or concepts?
We could probably minimize the latter two under experimental
conditions that prevent reflective processing, or by imaging
children or patients with certain kinds of brain damage. It
seems more problematic to dissociate emotional experience
from the functional emotion state, but that is a problem faced
also by all studies that want to isolate the neural correlates of
conscious experience (Block, 2015). A principal challenge will be
to construct ecologically meaningful situations that can induce
strong and well-characterized emotion states in the scanner environment.
Specific hypotheses about the functional roles that
particular emotions play, perhaps informed by work in evolutionary
psychology and ecology, would be needed to design the
best experiments. The ideal (difficult) project would design a
series of experiments across different species to study, for instance,
how the induction of fear across rodents, monkeys, and
humans might engage both overlapping neural systems as well
as components unique to a particular species. More realistically,
we could design human neuroimaging studies that not only
contrast different emotions (or attempt to discover distinguishable
subtypes of what we currently think of as one category of
emotion), but also vary the level of associated conceptual processing.
The recommendations for the affective neuroscientist
using fMRI are threefold: (i) partly disentangle the neural correlates
of emotions from all the other processing with which they
interact; (ii) carefully distinguish what aspect of “emotion” it is
you are investigating (states, experiences, concepts); (iii) construct
hypotheses derived from knowledge of the functional
features of emotions (Figure 2, which is not intended to be complete
by any means) and investigate these with the most sensitive
neuroimaging methods.
As an affective neuroscientist, I would want a framework for
investigating emotion that lets me investigate emotions not
only in healthy adult humans, but also in rats, in people who
cannot speak, and in people who are deluded about what emotion
they have. I would even want to be able to say something
to engineers who might want to construct robots that have
emotions. An operationalization of emotions as functional
states lets me do all of these, whereas a focus on emotion
R. Adolphs | 29
Downloadedfromhttps://academic.oup.com/scan/article-abstract/12/1/24/2624554byMasarykovaUniverzitauseron27February2020
experiences, or emotion concepts, does not. Again, all usages of
the word ‘emotion’ are interesting to study, and it may well
turn out that emotion states, emotion experiences and emotion
concepts are all related in interesting ways, and that they share
neural substrates. But that needs to be an empirical result, not
something we conflate in our research program at the start.
Acknowledgements
I thank Laura Harrison, Bob Spunt, Damian Stanley, Julien
Dubois, David Anderson, and Ajay Satpute for helpful comments
on the article.
Conflict of interest. None declared.
References
Darwin, C. (1872/1965). The Expression of the Emotions in Man and
Animals. Chicago: University of Chicago Press.
Scherer, K.R. (1994). Emotions serve to decouple stimulus and response.
In: Ekman, P., Davidson, R.J., editors. The Nature of
Emotion, pp. 127–30. New York: Oxford University Press.
Adolphs, R. (in press). Emotions are functional states that cause
feelings and behavior. In: Davidson, R.J., Shackman, A., Fox, A.,
Lapate, R., editors. The Nature of Emotion, 2nd edn. New York:
Oxford University Press.
Lazarus, R.S. (1991). Emotion and Adaptation. New York: Oxford
University Press.
Anderson, D.J., Adolphs, R. (2014). A framework for investigating
emotion across species. Cell, 157, 187–200.
Okon-Singer, H., Hendler, T., Pessoa, L., Shackman, A. (2015). The
neurobiology of emotion-cognition interactions: fundamental
questions and strategies for future research. Frontiers in Human
Neuroscience, 9, 58.
Pessoa, L. (2013). The Cognitive-Emotional Brain: From Interactions to
Integration. Cambridge, MA: MIT Press.
Simon, H.A. (1967). Motivational and emotional controls of cognition.
Psychological Review, 74, 29–39.
Tooby, J., Cosmides, L. (2008). The evolutionary psychology of
the emotions and their relationship to internal regulatory variables.
In: Lewis, M., Haviland-Jones, J.M., Barrett, L.F., editors.
Handbook of Emotions, pp. 114–37. New York: Guilford Press.
Sznycer, D., Tooby, J., Cosmides, L., Porat, R., Shalvi, S., Halperin,
E. (2016). Shame closely tracks the threat of devaluation by
others, even across cultures. Pnas, Available: www.pnas.org/
cgi/doi/10.1073/pnas.1514699113.
Ortony, A., Turner, T.J. (1990). What’s basic about basic emotions?.
Psychological Review, 97, 315–31.
Ekman, P. (1999). Basic emotions. In Dalgleish, T., Power, M., editors.
Handbook of Cognition and Emotion, pp. 45–60. Chichester,
UK: John Wiley.
Panksepp, J. (1998). Affective Neuroscience. New York: Oxford
University Press.
LeDoux, J. (2012). Rethinking the emotional brain. Neuron, 73,
653–76.
Caramazza, A. (1986). On drawing inferences about the structure
of normal cognitive systems from the analysis of patterns of
impaired performance: The case for single-patient studies.
Brain and Cognition, 5, 41–66.
Feinstein, J.S., Adolphs, R., Tranel, D. (2016). A tale of survival
from the world of patient S.M. In: Amaral, D.G., Adolphs, R. editors.
Living without an Amygdala. New York: Guilford Press.
Feinstein, J.S., Adolphs, R., Damasio, A., Tranel, D. (2011). The
human amygdala and the induction and experience of fear.
Current Biology, 21, 34–8.
Amaral, D.G., and Adolphs, R. eds. (2016). Living without an
Amygdala. New York: Guilford Press.
Feinstein, J.S., Buzza, C., Hurlemann, R., et al. (2013). Fear and
panic in humans with bilateral amygdala damage. Nature
Neuroscience, 16, 270–2.
Silva, B.A., Mattucci, C., Krzywkowski, P., et al. (2013).
Independent hypothalamic circuits for social and predator
fear. Nature Neuroscience, 16(12), 1731–3.
Berridge, K.C., Kringelbach, M.L. (2013). Neuroscience of affect:
brain mechanisms of pleasure and displeasure. Current Opinion
in Neurobiology, 23, 294–303.
Huth, A.G., de Heer, W.A., Griffiths, T.L., Theunissen, F.E.,
Gallant, J.L. (2016). Natural speech reveals the semantic maps
that tile human cerebral cortex. Nature, 532, 453–8.
Skerry, A.E., Saxe, R. (2015). Neural representations of emotion
are organized around abstract event features. Current Biology,
25, 10–1.
Tinbergen, N. (1950). The hierarchical organization of nervous
mechanisms underlying instinctive behaviour. In: Physiological
Mechanisms in Animal Behaviour, pp. 305–12, Volume IV. New
York, NY: Academic Press, Inc.
Lin, D., Boyle, M.P., Dollar, P., et al. (2011). Functional identification
of an aggression locus in the mouse hypothalamus.
Nature, 470, 221–6.
Damasio, A.R., Grabowski, T.J., Bechara, A., et al. (2000). Feeling
emotions: subcortical and cortical brain activity during the experience
of self-generated emotions. Nature Neuroscience, 3,
1049–56.
Mobbs, D., Petrovic, P., Marchant, J.L., et al. (2007). When fear is
near: threat imminence elicits prefrontal-periadqueductal
gray shifts in humans. Science, 317, 1079–83.
Mobbs, D., Yu, R., Rowe, J.B., Eich, H., FeldmanHall, O., Dalgleish,
T. (2010). Neural activity associated with monitoring the oscillating
threat value of a tarantula. PNAS Early Edition, Available:
www.pnas.org/cgi/doi/10.1073/pnas.1009076107.
Damasio, A. (2003). Looking for Spinoza: Joy, Sorrow, and the Feeling
Brain. Orlando, Florida: Harcourt, Inc.
Craig, A.D. (2008). Interoception and emotion: a neuroanatomical
perspective. In: Haviland-Jones, J.M., Lewis, M., FeldmanBarrett,
L., editors. Handbook of Emotions, 3rd edn, pp. 272–88.
New York: Guilford Press.
Lindquist, K.A., Wager, T.D., Kober, H., Bliss-Moreau, E., Feldman
Barrett, L. (2012). The brain basis of emotion: a meta-analytic
review. Behavioral and Brain Sciences, 35, 121–43.
Chikazoe, J., Lee, D.H., Kriegeskorte, N., Anderson, A.K. (2014).
Population coding of affect across stimuli, modalities, and individuals.
Nature Neuroscience, 17, 1114–22.
Saarim€aki, H., Gotsopoulos, A., J€a€askel€ainen, I.P., et al. (2015).
Discrete neural signatures of basic emotions. Cerebral Cortex,
doi: 10.1093/cercor/bhv086.
Wager, T.D., Kang, J., Johnson, T.D., Nichols, T.E., Satpute, A.B.,
Barrett, L.F. (2015). A Bayesian model of category-specific emotional
brain responses. PloS Computational Biology, 11(4),
e1004066.
Block, N. (2015). Consciousness, big science, and conceptual clarity.
In: Marcus, G., Freeman, J., editors. The Future of the Brain,
161–76. Princeton, NJ: Princeton University Press.
Phelps, E.A. (2006). Emotion and cognition: insights from studies
of the human amygdala. Annual Review of Psychology, 57, 27–53.
30 | Social Cognitive and Affective Neuroscience, 2017, Vol. 12, No. 1
Downloadedfromhttps://academic.oup.com/scan/article-abstract/12/1/24/2624554byMasarykovaUniverzitauseron27February2020
Adolphs, R., Tranel, D., Damasio, H., Damasio, A. (1994). Impaired
recognition of emotion in facial expressions following bilateral
damage to the human amygdala. Nature, 372, 669–72.
Spunt, R.P., Lieberman, M.D. (2012). An integrative model of the
neural systems supporting the comprehension of observed
emotional behavior. Neuroimage, 59, 3050–9.
Rapcsak, S.Z., Kaszniak, A.W., Rubens, A.B. (1989). Anomia
for facial expressions: evidence for a category specific
visual-verbal disconnection syndrome. Neuropsychologia, 27,
1031–41.
Barrett, L.F. (2006). Solving the emotion paradox: categorization
and the experience of emotion. Personality and Social Psychology
Review, 10, 20–46.
Wilson-Mendenhall, C.D., Barrett, L.F., Simmons, W.K., Barsalou,
L.W. (2011). Grounding emotion in situated conceptualization.
Neuropsychologia, 49, 1105–27.
R. Adolphs | 31
Downloadedfromhttps://academic.oup.com/scan/article-abstract/12/1/24/2624554byMasarykovaUniverzitauseron27February2020