In this post I will try to lay out my own account of what emotions are in a nutshell. The claim that emotions are embodied, roughly speaking, is that emotions involve bodily reactions and that these bodily reactions realize, or constitute, a kind of intelligent behavior, or interaction, with the environment. On my view different emotion types like fear, anger, or shame include different patterns of bodily reactions that evolved because they prepare the organism for different kinds of action. These bodily reactions are directly involved in the constitution of action-oriented representations. These representations establish a type of nonconceptual access to the world. They enable us to interact with our environment in an intelligent way. Through them we perceive objects as dangerous and as something that should be avoided in fear or as a rule-violation that we should compensate for in guilt.
The embodiment-hypothesis has its roots in the works of William James. It has been controversial ever since he articulated it near the end of the 19th century. In the book I discuss recent evidence from psychophysiology, and try to defend a minimal argument for embodiment, namely that bodily arousal is certainly not as random as authors like Nussbaum would have it. Rather, we find certain facial expressions that are typical of certain emotion types, such as sad faces or anger expressions, as well as certain bodily postures, such as the erected posture in pride, and also bodily reactions such as the increase of heart rate in fear and the reaction of the gastrointestinal system in disgust. I suggest that what we currently know about different kinds of arousal, bodily postures, and facial expressions in different emotion types makes it appear very plausible that most emotion types are associated with particular patterns of bodily reactions.
One complication in studying the bodily arousal involved in emotions stems from the fact that this arousal depends on many contextual factors ranging from the overall health of the organism to the situation in which the emotion occurs. This makes it not only hard to come up with comparable data, but also gives rise to the question of whether emotions can be individualized with reference to bodily arousal alone. The way an emotion sets up the organism for action in specific contexts certainly has to enter the picture somewhere.
Emotions, according to my view, are not only embodied but also action-oriented representations, that is, representations that not only represent something as a descriptive fact, but have a directive component at the same time. This is the kind of representation that Millikan (1993) calls “pushmi-pullyu” and Andy Clark (1997) calls “action-oriented.” I take it that the emotions’ being action-oriented is due to their being embodied. The embodied reactions involved in emotions are not randomly occurring arousal. They are well-adapted and skillful reactions that prepare the organism for action and thereby also realize an embodied action-oriented representation. I will say a bit more on what I take skills to be, since they are a central element of an embodied approach.
Skills are dispositions of agents. They enable living organisms to respond to certain situations in a spontaneous yet entrained way. Skills are abilities that presuppose a certain amount of training. Yet the kind of training in question is not necessarily guided by a trainer, and does not necessarily involve reflective thought and intention. It is a skillful ability to be able to hold one’s own head up and it takes a lot of training to learn how to hold one’s head up at a certain point in development, yet this acquiring this skill does not seem to require explicit guidance or reflective thought.
Likewise, I take emotions to be skillful abilities in this same sense. Emotions entail biologically established bodily reactions with a certain purpose, yet before single emotions can successfully fulfill their purposes many things need to be learned, and can be learned in different ways in different environments. Learned breathing patterns and body postures all contribute to the typical patterns with which we later on respond to emotion-relevant situations and thereby can vary depending on what the learning process looked like.
I also take emotions to be situated in a biologically and socially structured environment. There is a lot of literature on emotional scaffolding that I rely on here (see Colombetti & Kruger 2015 for a start). My focus lies on the ontological status of the intentional objects of emotions. I take these objects to be affordances and I develop an account of these affordances that fits into a naturalist, externalist framework. The naturalist framework excludes the commitment to supernatural ontological entities and the externalist framework demands that the extension – the class of objects an emotion correctly refers to – is defined externally. My point is that embodied approaches – if they want to claim that we can dispense with complex cognitive representations because we interact with a structured environment to which we are well adapted – need to say something about the structure of this environment: what kinds of affordances do we find there and what is their ontological status? I will answer this question in the next post on Wednesday.
Clark, A. (1997): Being There. Putting Brain, Body, and World Together Again. Cambridge (MA): MIT Press.
Colombetti, G.; Kruger, J. (2015): Scaffoldings of the Affective Mind. Philosophical Psychology, 28. 1157-1176.
Millikan, R. (1993): Pushmi-Pullyu-Representations. Philosophical Perspectives, 9. 185-200.
“emotions are embodied, roughly speaking, is that emotions involve bodily reactions and that these bodily reactions realize, or constitute, a kind of intelligent behavior, or interaction, with the environment.”
So, just to be clear, your view is that there are emotions (which are realized in the brain) and then there are reactions to those emotions (which are (somestimes?) realized in behavior.
To take an example, one’s anger might be realized in the brain, but then this brings about a bodily reaction, such as clenching one’s fists, that is realized in the body. Is this right?
thanks for pushing me on this. My formulation may have been unclear. What you suggest is not quite what I have in mind – and if it were I would not call it an embodied view, I guess. The claim for embodiment I want to make is stronger. The bodily reactions are supposed to play a constitutive role in the realization of the emotions aboutness. Take the example of anger: It is not only neuronal processing but also the increase of blood pressure, the adjustment of the muscles, the grim facial expression etc. that make you see something as an offense in anger. To be more precise: what you represent when angry, according to my view, is not just an offense, but rather an offense-to-fight, since I take emotions to be action-oriented representations that are about affordances. I take it that when being emotional our representations always include a descriptive and a directive component and I think that the bodily arousal explains the motivational component.
Thanks for the clarification, Rebekka.
My sense is that the embodied emotions view is the stronger view you describe. So, anger is realized by a whole brain-body-world complex. Anger doesn’t cause fist clenching; anger is constituted, in part, by fist clenching, right?
So, what if there were cases of individuals who were completely paralyzed, as by neuromuscular blockade. Would such persons not experience emotions?
Thanks in advance,
“Anger doesn’t cause fist clenching; anger is constituted in part by fist-clenching, right?”
partly yes… but not quite. My claim is that a pattern of skillful bodily reactions plays a constitutive role in the realization of an emotion. Fist clenching can be a part of that pattern but is not a necessary ingredient. Also fist clenching is a typical output of an emotion such as facial expressions. It is indeed part of an embodied view to claim that input and output are not clearly separated, and that expressions and postures often are part of what causes an emotion (see e.g. Laird 2007). But the more central ingredients of anger are change in heart rate, blood pressure, release of hormones etc.
Depending on the kind of fist clenching (reflexive or voluntarily? early or late?) one might also want to say that it is indeed rather a follow up behavior caused by the emotion than a constitutive part, without thereby undermining the claim that the emotion is constituted by a pattern of bodily reactions.
Now, with regard to your question: I am indeed committed to the claim that in the complete absence of skillful bodily reactions people should not experience emotions. This is of course a controversial claim, and I’m happy to modify it if convincing evidence suggests to do so. Yet the devil is in the details and the question is what would count as an interesting case of counterevidence. I have doubts about your example of patients that experience fear during a neuromuscular blockade, since I suppose that although these people cannot move, they still be aroused – increase in heartrate etc. My prediction would be though that the experience of fear changes and also decreases in intensity, when there is no bodily feedback. This seems to be the case in people with spinal cord injuries and also in patients whose facial expressions are limited due to botox injections. So I must admit that I’m surprised that studies on people experiencing fear during neuromuscular blockade do not report a changed or diminished experience of fear. But I don’t take that to be decisive evidence against my view.
There is some discussion of this point in the emotion literature as you probably know. Hohmann (1966) report subdued emotion in patients with spinal cord injuries with the exception of sadness. They also found that the more degraded the bodily feedback, the greater the reduction in affect.
Chwalisz et al 1988 by contrast found reports of intense emotion experience in spinal patients. However this could be explained by information from the viscera carried by the vagus nerve which project to the brain-stem, as Rebekka also suggests below.
The evidence doesn’t decide the issue in favour of the disembodied theory in any case.
In truth, I’ve not read much on embodied emotions, since it seems to me to raise all the same issues as does embodied cognition. (I did read a paper by Jan Slaby, another by Sven Walter, and the first part of a draft of Colombetti’s book.)
I think the cases you cite probably show the causal influence of the body on emotions, but not a constitutive role for the body in emotions. What would seem to me to indicate a constitutive role would be that the loss of the body part prevents there being a particular emotion.
In keeping with that thought, I would predict that there are going to be brain lesions that prevent there being certain emotions. That’s because the brain is the supervenience base of the emotions. But, that’s just a prediction. I’ve not read the literature, but I would speculate that if one were to read through it one would find such a study.
While I think it can be hard to disentangle causation from constitution, there is no reason to despair.
Well, if you start to read neuroscientific studies looking for the supervenience base for emotions in the brain, like the meta-study on fear responses in the amygdala I quoted above, there soon might be reason to despair…
Perhaps, but one challenge with brain lesion studies is the high variability in localization and size of brain regions. (Area V1, for example, which is relatively easy to identify anatomically, varies three-fold in size.) And, with humans, it’s unethical to induce controlled lesions, so, it is generally pretty difficult to be confident that natural lesions focus on just the putative supervenience base of an emotion or any cognitive function. And, there is the possibility of compensatory adjustments and the recovery of function in adjacent areas. There must be other factors that I am omitting here. So, I don’t find it all that surprising that meta-studies have troubles. There is not yet enough experimental control. The prediction still stands, though experimental and clinical technique can be challenging.
“My prediction would be though that the experience of fear changes and also decreases in intensity, when there is no bodily feedback.”
This sounds to me like a disembodied view. There are emotions that causally influence, say, heart rate, then heart rate causally feeds back to influence the emotion. The embodied approach seems to have this story available. How does the story go if the emotion is constituted by the heart rate?
Thanks in advance,
Ken, I don’t think Rebekka’s position has to be read in that way. It might be that she’s allowing here that some (weak) emotions don’t involve bodily reaction and thus that emotion isn’t *necessarily* bodily or embodied, but still in the case where bodily feedback is present that feedback could be a constitutive part of the (strong) emotional reaction, and not just a causally contributing factor. Right? Not that it is clear to me how to adjudicate the difference …
“It might be that she’s allowing here that some (weak) emotions don’t involve bodily reaction and thus that emotion isn’t *necessarily* bodily or embodied”
One common reaction to paralysis cases is to change quantifiers, so it is not that emotions are embodied, but that some emotions are embodied. That avoids the simple counterexample move, but it points up a theoretical vacuum. If only some emotions are embodied, which ones? Why those? What is the evidence for those emotions being embodied, but the others not?
“in the case where bodily feedback is present that feedback could be a constitutive part of the (strong) emotional reaction, and not just a causally contributing factor”
But, the disembodied view accepts that bodily feedback is part of the emotional reaction. What the disembodied view rejects is that bodily feedback is part of the emotion. (See my comment of MARCH 23, 2016 AT 8:44 AM and Rebekka’s reply.)
There is a lot of doubt out there about how to settle what constitutes an emotion versus what causally influences an emotion. But, much of that derives, I think, from lack of clarity about what one is talking about in talking about emotions. If you can’t pin down what you mean by an emotion, it’s going to be hard to pin down what constitutes it. So, for example, one needs to be careful to distinguish emotions from reactions to emotions.
Hi Ken, these are good questions, and I don’t have ready answers to them — though I know there are other philosophers who claim they do. I was just pointing out that Rebekka’s concession that there could be some disembodied emotions didn’t require saying that some emotions are disembodied. This was meant to be a clarifying point, and I think it’s important, even if — as I agree — you’re right that proponents of embodied views could be clearer in their own ways.
Is there any chance you have references to places where folks say that some emotions are embodied and some are not and how they divide them?
Thanks in advance,
Oh, I don’t know where (or if) this is discussed in the emotions literature, beyond what Rebekka has said here. I was thinking of related claims in the wider literature on embodied cognition and consciousness.
Ok. I guess I have just missed them. Too much to read!
Hi Ken, sorry for the easter-delay, and thanks John for jumping in.
First of all: I don’t want to claim that there are some (weak) disembodied emotions. I think that all emotions are embodied. The passage you quote, Ken, was just not well put (to say the least). What I meant is *less* bodily feedback, not *no* bodily feedback. Sorry for causing so much confusion.
I’m not aware of anybody in the literature who explicitly changes quantifiers in the way you suggest, Ken (Colombetti for example says what I tried to say above, too: that it is not clear whether there is no feedback at al in the paralysis cases. There might still be visceral feedback). If the claim is that bodily arousal plays a constitutive role for the emotions aboutness, I don’t think that it is a very attractive option to call some emotions disembodied. You would end up with two different emotion categories and for the disembodied class you would have to explain how the emotions in this class can be about something at all. I would rather say that somebody might have judgments such as “What an achievement!” or “That’s really dangerous” and that these might be disembodied and that from a phenomenological perspective these might be hard to tell from mild cases of pride or fear sometimes.
Ken, you say: “So, for example, one needs to be careful to distinguish emotions from reactions to emotions.” That is right and for an embodied account that becomes a fuzzy affair. But if your suggestion is to distinguish emotion and emotional reaction such that there are clearly separated (sensory) inputs that cause an emotion, a neuronal program that receives them and then causes well coordinated bodily/behavioral responses (as e.g. Griffiths would suggest for basic emotions) then I wonder whether you would say that the evidence for this view is any better than for an embodied view. As far as I know there is no evidence for a neuronal program that receives input and guides output in the way Griffiths would suggest, but there is evidence that input and output in the case of emotions cannot clearly be separated (e.g. Laird 2007). I certainly don’t think that there is enough evidence so far to make a definitive argument for any approach. But I do think that embodied dynamical views such as my own or Colombetti’s (2014) by giving bodily feedback a constitutive role can handle the existing evidence in a more convincing way than classical input-output approaches.
Thanks Rebekka, this is really helpful!
“I’m not aware of anybody in the literature who explicitly changes quantifiers in the way you suggest, Ken ”
It’s pretty common in the extended/embodied cognition literature to maintain only that *some* cognition is extended/embodied. I think Clark and Rowlands hold that view. Noe, I think, changes quantifers regarding perception, when it comes to dreams. Rob Wilson changes quantifier to handle the case of paralysis. So, it seems merely to be an accident that no one in the embodied emotions lit has changed quantifiers.
Does your book have a discussion of the difference between causation and constitution?
In my book, you could show constitution if you were not to have an emotion with the loss of, say, the heart. If having a heart (or a heart’s beating, or whatever) is part of what is it to have emotion X, then having your heart removed should prevent you from having emotion X, rather than modulating the emotion X. Of course, say, botox might causally influence your emotions, but that is fine with the disembodied story. What one would seem to need is that with botox, there is no emotion X. Something like that.
now that you mention the dream-cases Noe struggles with, some cases of changed quantifiers in the emotion debate come to mind (if your notion of what counts as an embodied approach is not to strict that is): Damasio (1999) rediscovers James, describes emotions as read-outs of bodily arousal and then introduces neuronal as-if-loops, allowing that certain emotions do not involve actual arousal but rather simulate it on a neuronal level. Prinz (2004) adapts this view.
The important point here is that these are piggyback-cases (which, if I remember correctly, is true for Noe as well). The as-if emotions depend on the normal cases and the whole body-system that realizes them. So because these are piggyback-cases you don’t have to come up with a whole new story about the intentionality of these states.
So, the as-if-loop could serve as a response to the paralysis-cases. I just don’t see that it is even necessary to go this far, since I’m not convinced that the paralysis cases offer sufficient evidence of there being no bodily feedback at all.
The causal/constitution-debate is not a central issue in my book, because I don’t see much room for development/new decisive arguments in the debate as it stands. So I don’t have much hope to convince somebody like you, given the current state of the sciences 🙂 The hope of the book rather is to convince people who think that emotions are cognitively very complex states with a normative dimension that this dimension can be accounted for in an embodied framework. Therefore the parts of the book that I do consider most interesting and original are those on the emotions intentionality and their being about affordances.
You find what I have to say on the causal/constitution-problem in the 2. Chapter though. From what I say there, it should be clear that I take *patterns* of bodily arousal to be constitutive for emotions. Neither heart rate nor facial expressions are necessary parts of these patterns and this is why removing your heart would not necessarily prevent you from having an emotion.
“since I’m not convinced that the paralysis cases offer sufficient evidence of there being no bodily feedback at all.”
Well, the paralysis cases display the logic involved in testing the hypothesis: Knock out the bodily component and see if the emotion exists. So, suppose it’s the heart. Remove someone’s heart and keep them alive on a machine. Such a person would then not have an emotion which is constituted by a change in heart rate. (Or maybe do it with someone with a pacemaker.)
I haven’t read your book so I’m just giving a simplistic cartoon reply, but all the examples I’ve heard of so far in both the extended/embodied cognition literature and e/e emotion literature just report modulation of emotions, rather than absence of emotions.
“The causal/constitution-debate is not a central issue in my book” Well, it seems to me that there are really just about two issues. We need to be clear about what we are taking about with “emotion” and we need to be clear about causation versus constitution.
“So I don’t have much hope to convince somebody like you, given the current state of the sciences ”
Well, the paralysis cases could have worked for the embodied approach, but they don’t. I could have been convinced. And, I think that issues of causation versus constitution are regularly solved. It’s pretty clear that a nail causes a flat tire; it does not constitute a flat tire. High consumption of alcohol might cause depression, but it does not constitution depression. These seem to be unproblematic, so why is it so problematic wth regard to emotion? (I think it is lack of clarity of what an emotion is., just as there is a problem with the “mark of the cognitive”. Note that Rowlands, 2009, “Extended cognition and the mark of the cognitive” also has this view.
Well, I’d be happy if causal/constitution questions were as easy to answer as you suggest. But even if we all agree that alcohol does not constitute a depression, I’m not aware of any emerging consensus in the philosophy of psychiatry on what actually does constitute a depression. Same in the case of emotions.
You are right that empirical evidence so far only shows that there is modulation in emotions that covaries with modulation on the physiological level. Given that the patterns of bodily reactions we are talking about are complex and not easy to measure I think that is already something.
Compare studies on the neural realization of emotions. One of the most common claims is that the amygdala plays a central role in fear processing. It would probably be the best candidate for a constitutive component of fear in the brain. Yet a meta-analyses (Phan et al. 2002) reported that only 60% of studies involving fear showed increased activation in the amygdala. So you don’t even have to remove the amygdala to cast doubt on the claim that it could be a constitutive element of fear in the sense that you seem to be looking for. You might conclude from this that we just don’t have the empirical evidence to make convincing constituion-claims so far (which then is not only a problem of embodied accounts but of brainbound accounts as well) or you might consider that emotions are realized by dynamical patterns that develop and change over time and display a high degree of multiple realizability – and if you believe the latter you will not expect that the consequence of removing the heart or the amygdala will be that fear ceases to exist but rather that the emotion will be…. modulated.
“Well, I’d be happy if causal/constitution questions were as easy to answer as you suggest. But even if we all agree that alcohol does not constitute a depression, I’m not aware of any emerging consensus in the philosophy of psychiatry on what actually does constitute a depression. Same in the case of emotions.”
It’s, of course, true that there are going to be philosophical issues regarding causation and constitution. These are active areas of philosophical research, but the distinction is pretty central to the embodiment movement and the embodiment movement has been surprisingly indifferent to the issue.
As I see it, there is the “What is cognition?” “What is an emotion?” and “What is depression?” sort of question (The “mark of the cognitive”). These are roughly equivalent to “What constitutes cognition?” “What constitutes an emotion?” and “What constitutes depression?” on one reading of the word “constitutes”. It is a kind of old-fashioned philosophical question. Once you get clear about what you mean by “cognition”, then matters are a lot simpler. So, for example, Chemero uses “cognition” to mean what cognitivists call “behavior.” Well, ok, then clearly *that* sort of “cognition” is extended and embodied. Once you are clear on that then you can wonder about why we should use “cognition” this way or why cognition might really be behavior.
So, I don’t share the skepticism about the causation/constitution distinction. I think that problems with that distinction tend to be conflated with relatively traditional problems about “What is X?”
You wrote above:
“Well, the paralysis cases display the logic involved in testing the hypothesis: Knock out the bodily component and see if the emotion exists.”
The trouble is that the paralysis cases that I know of don’t deliver this result. They don’t constitute cases in which there is no bodily feedback of the kinds that might play a functional role in affective processing.
“The trouble is that the paralysis cases that I know of don’t deliver this result. They don’t constitute cases in which there is no bodily feedback of the kinds that might play a functional role in affective processing.”
You are exactly right, Julian. The paralysis cases knock out certain bodily contributions, but not others. But, that’s why I said that the “logic” of the paralysis cases still holds: “the paralysis cases display the logic involved in testing the hypothesis: Knock out the bodily component and see if the emotion exists.”
So, following up, what one might do is remove a patient’s heart while the patient is not under general anesthesia, keeping the blood pumping by some artificial means. If the heart rate is constitutive of the emotion, then would expect emotional deficits, as in, that person has no fear. There can be ethical and practical issues with this, but it’s not a conceptual problem.
One has to look at scientific evidence to sort this out, but I think the disembodied prediction is that brain lesions will lead to emotional gaps, where “cardiac lesions” will not lead to emotional gaps. The embodied prediction, I think, is that both brain lesions and “cardiac lesions” will lead to emotional gaps.
FYI, as an aside, Miriam Kyselo and Ezequiel DiPaolo (Locked-in syndrome: a challenge for embodied cognitive science) have a paper where they seem to accept that paralysis cases are a problem for the view that the body plays a constitutive role in cognition.
Their “solution” is to propose another sense of the word “constitution”. I don’t think their “solution” helps, but they respect the problem.
Just quickly: The main argument that Miriam and Ezequiel offer is actually very similar to what I have been trying to say in earlier posts. They say:
“From an enactive perspective a LIS patient is bodily active because embodiment is more than neuro-muscular movement.”
I’ve been trying to say that I would expect emotional experience to change and probably reduce in paralysis cases, but I would not expect it to disappear, since there still is a lot of bodily feedback that does not depend on muscle-movement.
Miriam and Ezequiel say basically the same about patients with LIS and their cognitive abilities. I Would only disagree with their claim that extended functionalists cannot account for these cases.
Ken, I will be offline again for the weekend. But thank you very much for your interseting questions!
Yes, their positive account coincides with your account, but they do get that LIS cases are a problem for a version of Noe’s view that holds that action is bodily movement.
Anyway, thanks for putting up with my dogged questioning.