Thanks to John and the team for letting me take over the Brains Blog for the week. Over the next five days I’m going to summarise some of the key ideas in my new book The Emotional Mind: A control theory of affective states (Cambridge University Press).
The book grew out an attempt to give an original account of the emotions. As anyone who has researched this area can tell you, the more you explore the emotions, the more you realise how complex they are. They connect to all sorts of issues about bodily sensations, actions, values, reasoning, character and social interactions. Ultimately, I don’t think a proper account of the emotions is possible without understanding how they are placed within the mind as a whole. This is what my book tries to provide. As a result, I end up with a theory of mental architecture, where the mind is structured around fundamental life-sustaining processes that deploy ever more elaborate forms of cognition to serve their aims. I will say more about this architecture on Friday. Today I want to start with the theory of emotions at the heart of the book.
First some brief background: The history of the emotions debate since the time of William James has been dominated by a debate between the traditional cognitivist position- that emotions are a species of evaluative judgement, and the non-cognitive or somaticist position- that emotions are patterns of bodily sensations. As far as I can tell, everyone these days is searching for the ideal middle ground where evaluative judgement and bodily responses are perfectly synthesized. It is not enough to say that emotions combine bodily responses and judgement. The key adaptive innovation of emotions will lie in how exactly these two features are connected.
In recent times the perceptual theory of emotions has presented a particularly strong synthesis. As Prinz (2004) articulates it, the feelings of bodily changes themselves represent the intentional contents of emotions. However, there have been a number of attacks on this theory, and in my opinion, the perceptual theory is dead. Michael Brady (2013) has a book length attack on the epistemic analogy with perception. It is also clear that emotions are routinely mediated by other perceptual or mental states (like imagining, memory, and the interoception of bodily responses) in a way that just doesn’t fit the profile of perception. See Salmela (2011), Deonna & Teroni (2012: Ch. 6), Barlassina & Newen (2013) for further attacks, plus a few in my own book.
A new kind of emotion theory has tried to move beyond the perceptual theory by arguing that bodily responses should be understood less as the representational content of an emotional state, and more like its mode or attitude (see e.g. Deonna and Teroni 2012, also Hufendiek 2016 to some extent). That is, we don’t represent danger (or loss, or gain); we represent in a threatened, or protective or welcoming mode, by responding in the ways characteristic of fear, sadness of happiness respectively. I similarly regard the evaluative aspect of emotions not as something that we represent, but as captured in a valent manner of representing. That is, we represent in a valent way, by being disposed either to increase the presence of the intentional object (positive), or decrease its presence (negative). Tomorrow I will also argue that this valent treatment of the object is vital for fixing the referent of the mental state, thus supporting an embodied theory of mental content.
Where my theory differs from that of Deonna & Teroni and Hufendiek is that I still attribute a constitutive role to cognitive representation in helping us to grasp the object of the emotion and to trigger the appropriate responses. I share this view with traditional appraisal theories in psychology. A particular innovation of mine is to describe the characteristic form that these representations take. I claim that all emotions represent a contrast between the current state of affairs and some other state not in the here and now. For example, fear involves a representation of an upcoming harm (a temporal contrast), jealousy represents someone else getting a good thing (a social contrast), and regret represents a bad thing that might not have happened (a modal contrast). I call these situated concerns, to distinguish them from way that pains and pleasures track immediate harms and benefits. The representation of contrasts makes emotions more sophisticated than pains and pleasures because the subject is now oriented towards the wider context, and can accordingly serve his or her interests more in a contextually sensitive way. Overall, my slogan definition is that emotions are ‘valent representations of situated concerns’.
Another way my theory differs from somaticist theories is that while I think bodily responses play a necessary role in emotions, I do not think the feelings of those changes are necessary for emotions. I think bodily feelings play supplementary role in capturing a distinct, though complementary sort of intentional content which I call emotional bodily feeling (see also my 2017 paper, which is an earlier version of chapter 4 of my book). Indeed my account of emotional bodily feelings resembles Hufendiek’s account of emotions, because we both take bodily feelings to represent the capacity of the body to deal with the situation. However, I believe that somaticist theories are overly focusing on this aspect of ordinary emotional experience and failing to properly recognize the work that quite sophisticated cognitive representation is doing in capturing intentional contents, prior to the experience of bodily responses.
Overall, my theory is slightly more cognitive in flavour compared to some recent theories, yet bodily responses are still playing a constitutive role in fixing what the emotions are about, and the evaluative manner the subject regards these objects. Tomorrow I will give a deeper account of just what a valent representation is and why we should recognize it as a fundamental sort of mental state.
References
Barlassina, L. & Newen, A. (2013). The Role of Bodily Perception in Emotion: In Defense of an Impure Somatic Theory. Philosophy and Phenomenological Research, 89(3), 637-678.
Brady, M. S. (2013). Emotional Insight: The Epistemic Role of Emotional Experience. Oxford: Oxford University Press.
Cochrane, T. (2017). The double intentionality of emotional experience. European Journal of Philosophy, 25(4), 1454-1475.
Deonna, J. A. & Teroni, F. (2012). The Emotions: A philosophical introduction. Oxford: Routledge.
Hufendiek, R. (2016). Embodied Emotions: A Naturalist Approach to a Normative Phenomenon. New York: Routledge.
Prinz, J. (2004). Gut Reactions: A Perceptual Theory of Emotion. Oxford: Oxford University Press.
Salmela, M. (2011). Can Emotion Be Modelled on Perception? Dialectica, 65(1), 1-29.
I may have missed something here. Your approach seems to focus on the form and content of emotions. It doesn’t seem to address the primary question of function – why are there emotions in the first place, when computers/robots which provide the current paradigm for cognitive science don’t have or need emotions at all? More particularly – 1) why is the living mind bicameral, with a conscious self and mind system that feels the pains of emotions, when computers/robots are unicameral, and without feelings ? 2) why is the conscious self pressured to act, by an enormous array of emotions, which are all to some extent painful, pressuring the self to act to remove them? why does the self by extension use such an extraordinary variety of methods to deal with and moderate emotions, including mood-altering substances and strategies? – computers/robots aren’t and don’t have to be pressured/pained by emotions; have only to be instructed to act 3) why is the conscious self always pressured by *conflicting* emotions, pulling/pressuring it in *opposite* directions – e.g. to eat//resist the sweets? – again, computers/robots are not subject to conflicts of any kind. Overall, the entire conscious self and mind system with its continuous human drama of conflicting emotions (a drama entirely missing from the sciences!) makes no sense at all within present cognitive paradigms, and an account of emotions must surely explain it.
PS One should add a further question – 4) why can unicameral computers/robots not work at all without a human conscious self and mind operating them? why has technology failed to take the conscious self “out of the loop”? why are all machines, when considered as wholes, still parts of bicameral systems?
P.P.S. I should have added that the fact that the human drama, which fills our every waking minute, and is the stuff of the dramatic arts, religions, and all our media, is entirely overlooked by the sciences, is in large measure due to science having no way at all of explaining it within science’s present worldview.
Hi Michael. The main function of emotions is, of course, to keep us alive. When I say “the subject is now oriented towards the wider context, and can accordingly serve his or her interests more in a contextually sensitive way”- I take it as read that one of our main interests is to stay alive.
Since computers/robots don’t care about staying alive, their cognition is not oriented around this. And in my opinion, they will never really seem conscious to us until they do care about staying alive.
Frankly, that’s general waffle. “Staying alive”? What does, say, being conflicted about whether to have a doughnut or two or stick to the diet, have to do with staying alive? Why is a soldier being tortured, conflicted between talking and ceasing the pain OTOH and possibly resisting and dying for the cause OTOH? There are a good range of emotions/urges, as per the last example, that pressure people to *die* and not stay alive. Why?
I issued a set of absolutely central challenges/problems about the function of emotions. If you don’t want to deal with them – much as science refuses to deal with the human drama – that’s your choice. “Staying alive” doesn’t answer any of them.
Hi Michael. Please do not expect blog comments to be particularly careful. However, to give you a slightly more careful response- what I say is that emotions serve our interests or concerns. We can have lots of different concerns, some of which conflict with each other, as per your examples. Serving some concerns (e.g. protecting your loved ones) can also end up leading to your death, and more generally, there are lots of concerns the satisfaction of which is not healthy.
What I thought your original question was getting at was the biological history of the emotional capacity- which as per a standard evolutionary story will just be ultimately to serve survival/reproduction. Moreover, for every interest/concern we have, there should be some (probably convoluted) story to tell about how that concern has become associated with a basic homeostatic regulating function. That this process can ultimately lead to unhealthy emotions is not surprising, since association is not especially rational.
Thanks for your comments, but please watch your tone as you’re coming across a bit rude and you seem to be bearing some kind of grudge against ‘science’.
No, my questions/problems are about the function not the evolution of emotions. They can all be re-expressed as : “why would you give a robot a bicameral mind with human/animal-like emotions?” (when current robots don’t have them). That has to be dealt with.
I have a “grudge” against science, as Francis Crick had a “grudge.” He said it was “ridiculous” that science didn’t study consciousness, and helped usher in consciousness science/studies in the 1990’s. That’s not a grudge, that is a deeply principled objection to science being deeply unscientific and shirking its duty.
While science has finally, after 90-odd per cent of its history, graciously started to study consciousness, it still isn’t really studying the conscious self, which would entail studying the self’s inner debates including its struggles with conflicting emotions. When science began, it in effect ceded that territory to Shakespeare (who was mainly responsible for introducing the inner monologue c 1600) and a million dramatists since.
That science still isn’t studying the conscious self and its drama isn’t just “ridiculous”, it’s scientifically obscene – like studying Hamlet without the Prince, the Apple organization without Steve Jobs/Tim Cook – the vehicle without the driver. And you can’t study emotions scientifically without starting from the conflicted self.
I’m going to have one more go at responding to you. We would need to give a robot emotions if we wanted them to protect themselves against potential harms, or pursue potential opportunities- i.e. to be self-interested. I’m not sure we want this though. I can’t see it being compatible with Azimov’s three laws.
Meanwhile, people working in the philosophy of emotions certainly care about emotional conflict (i.e. the traditional debate about reason v emotion) and akrasia. I’m not going to talk about it much here but I will address it briefly on a post on the CUP blog EDIT: the CUP post is here.
This finally, is from page 1 of my book:
“To grasp this complexity is the principal aim of this book. But my strategy is not that of a novelist offering nuanced depictions of lived experience. I aim to proceed systematically; to divide the mind up into relatively discrete levels and proceed from bottom to the top, pointing out the key components and structural relationships.”
Disagree. If you want robots to protect themselves, you simply instruct them. No need for emotions. No need, above all to torture them. No need for such an extraordinarily elaborate emotional system.
Emotional conflict means conflicting urges/emotions – not emotion vs reason.
My impression is you haven’t really grasped the profound nature of the challenges I identified – that’s not to be critical; it’s simply that if you haven’t thought long and hard about these matters, you can’t really deal with them off the cuff.
I won’t go into the great detail needed here or fully crystallise their differences, but the broad nature of the solution to the challenges, is that the two minds, conscious and unconscious, are fundamentally different and *opposite* systems, dealing with fundamentally opposite kinds of problems. Kahneman’s “thinking fast” vs “thinking slow” doesn’t really cut the profound opposition of these systems. The unconscious mind, like all machines to date, doesn’t need emotions because it’s dealing with the “dumb”, easy problems, for which it is already scripted and has the answers. The conscious mind, which is the smart system, is dealing with the smart, hard problems, for which there are no right answers or definite methods of solution, and which, as Damasio illustrates, a mind can think about forever. That’s why this system needs to be pressured to take decisions, and not go on forever. *But* the system cannot be pressured in a single direction – like protecting itself, or “staying alive”. It has to be pressured in conflicting directions, precisely because there are no right answers here. The conscious self is there to take the hard (rather than the easy) decisions. The unconscious mind thinks automatically, smoothly, effortlessly about its easy problems. The conscious mind thinks deliberately, hesitantly, struggling with conflicting emotions about its hard problems. It’s actually a neat division of labour.
Fully understanding the two systems means transforming our ideas of human nature and intelligence, (which is far, far more than the mere “rational intelligence” of a “rational animal”).
Hi Tom,
Really interesting stuff! I’m teaching a course on emotions and cognition right now, and I might recommend these posts to my students.
Quick question on the following point: “That is, we represent in a valent way, by being disposed either to increase the presence of the intentional object (positive), or decrease its presence (negative).”
I wonder how this point applies to something like grief. First, what is the object of such an emotion. I would have thought it is something like, e.g. Ann’s (some particular) death [Side comment: Is this a temporal contrast? I wasn’t quite sure from the above what a contrast was, and I worried that we could introduce a contrast for every kind of content…maybe this will become clearer later]. Anyhow, my first worry is that this is not the sort of thing that can be increased/decreased. A natural move is to say that the object is the representation of the person’s death, and that what an emotion’s being negative consists in is in drawing our attention away from the thing (something like this seems plausible with respect to what’s negative about e.g. disgust); but it seems actually as though grief *draws* our attention to the object in question.
Feel free to delay replying if this is something that will be easier to respond to once you have more material on the table.
Looking forward to your other posts!
Best,
Juan
Thanks Juan.
Grief is one of the most complicated emotional processes there is, so I’ll say something about sadness first, and then briefly note the main additional complication about grief. This will be helpful in giving a fuller sense of how my emotion model works.
Sadness is already a fairly complicated case, since it requires at least 3 control layers. First we have the underlying concern- which is a positively valent attraction to something (e.g. your beloved). Basic attractions and aversions do not require contrast representations. When the ‘level’ of the attractant is low, a response is triggered to increase the level. (This will be explained more in Tuesday’s post on valent representation).
Second layer: We represent the attractant as lost (yes this is a temporal contrast). This representation automatically triggers a response to reduce the presence of the loss (so it’s a negatively valent representation in that sense) by restoring the presence of the lost thing.
So far, this response is common to both anger and sadness, so we need a third layer
to differentiate anger from sadness. The third layer then is where the initial response is represented as inadequate to restore the loss- leading to a secondary response to inhibit the first, and engage compensatory responses instead (so note that a representation of the bodily response is necessary here in a way that is not necessary for all emotions). That is, sadness is overall the valent representation of being unable to restore a loss.
The major complication of grief is that the underlying concern attaches to a unique individual, i.e. love. Because the attachment is to a unique individual, this makes it impossible for the usual response to either restore or compensate the loss. Thus a slow process is required where the control system has to gradually learn to no longer seek the presence of the loved one.
I discuss the case of grief in chapter 7 of my book. You may be aware that Peter Goldie has a paper on this emotion, which he uses to defend his narrative model, because he thinks there is no essential emotional response or representation (instead we have different stages of grief). However, I argue that we can treat grief as fundamentally a control process in the manner sketched above.
Thank you for this really helpful reply, Tom. I have a much better sense of your account from this. I have a couple of follow ups:
1. First, just a clarification: It looks as though for a lot of emotions we are going to have a mix of diminishing and increasing dispositions. So what on this account explains the overall valence of an emotion? Is it the valence that the highest representation has?
2. What reasons do we have for positing (quite generally) the second layer for sadness? i.e. Why think that anyone who feels sad is disposed to reduce the loss? In cases where it is clear to the person that nothing can be done, it seems implausible to me to posit this layer.
3. I wasn’t clear if what differentiated sadness from anger on your account was that (a) anger had a different third layer, or (b) that it had no such third layer. Either way, it seems to me that we can have a representation of the sort invoked while feeling angry. E.g. I may get angry every time I think about my inadequate response to an insult several years ago about which I can’t do anything anymore.
4. But maybe the difference between anger and sadness is not coming from the representation of inadequacy, but rather from the particular response triggered (“to inhibit the first, and engage compensatory responses instead”). But the first response in particular doesn’t seem necessary for happiness. And I’m still not sure how it differentiates it from anger, but maybe I’m not understanding what these responses involve.
Best,
Juan
p.s. I enjoyed your second post. My main question about it was why you hold the following: “To require both percepts and desires to appear simultaneously is implausibly complex as an account to how content naturalistically emerges.”? Why isn’t the argument here like: ‘It would be too complex for two legs to develop at once, so we must suppose that the most basic case is that of a monopede’?
Hi Juan- these comments are a bit more challenging!
1. I don’t have a rule for deciding overall valence. On many occasions it will just be mixed in the way the sadness example was. Note that other episodes we categorise under sadness have an avoidant object at the root- for instance, you wanted to avoid getting hit, but you got hit, and now you can’t restore your standing by say, hitting back. So I should have been more careful in the previous post when I said that ‘sadness overall is…’ (it’s not a definition taken from the book). Still, I’m happy to say that the key common feature in sadness will be this lack of capability at the third layer. This is analogous to what psychologists like Scherer call ‘coping potential’ (I have a section in Chapter five on this).
2. The main reason for the initial restoration response is again specific to cases where there’s an underlying attractant. It’s just an a priori feature of being an attractant that you are disposed to get it if you can. Also I don’t think you can represent a lack of capability without in some way modeling your capability- even if it’s simulated rather than enacted. If you’ve seen the film Brokeback Mountain, you may recall the very moving scene at the end where Ennis finds Jack’s shirt and embraces it. He wants Jack, so he grasps at this token, but he’s also keenly aware that he cannot get him back.
3. To clarify- the key difference with anger is that you *do* feel powerful enough to meet the challenge, and you gear yourself precisely to do this- to get back the lost thing, or to remove barriers to getting it. However anger can burn itself out if, by trying, you find you can do absolutely nothing about it- in which case I would expect the response to switch to sadness. I’m not sure about the example you offered, but the best way I can make sense of it is where you’re kicking yourself for a mistake. If so, your anger is directed at yourself, which you represent as capable of not making such mistakes. Thus you do have the capability for restoration in the sense that you can make sure you’re really a better person than that. Complicated case!
4. I’m no so clear what you’re getting at here, but I think the above comments might have clarified what I mean.
About the valent representation argument, I’m not sure if you’re objecting or what. I guess yes? two fully developed legs (as opposed to say, a bumpy undercarriage) does seem too complex as an initial development. Perhaps you could add a comment on that post if I’m not getting it.
Really interesting and, to me, phenomenologically plausible. I have nothing to say about the big picture, but I would like to suggest a more careful characterization of jealousy. I think your OP describes envy, not jealousy.
To my mind, and using your vocabulary, jealousy arises as a representation of someone else’s getting a good thing that is or used to be yours (in some sense). In contrast, envy arises as a representation of someone else’s getting a good thing that has never been yours (in some sense). To contrast them in brief: jealously represents a loss, and envy represents a lack. One might be envious when one’s colleague gets a promotion, but jealous when one’s lover leaves one for that colleague.
Of course, I would not be (and never am) surprised to find out I’m wrong.
I think that’s a pretty good distinction, so long as we make clear that in jealousy, we might not have actually possessed the good thing, but only take ourselves to have a right to it. So I could feel jealous about a colleague’s promotion if I thought I was also qualified and I’ve been passed over. Meanwhile for envy it sounds right to say that it’s a lack where I have no such expectation but only maybe an aspiration. It’s close to admiration in that sense.
Nice. I see the distinction between (a) having actually possessed the good thing and (b) having not possessed but felt owed the good thing, and I see how both could be represented in jealousy.
I was thinking of drawing a similar distinction within jealousy between (a) having actually possessed the good thing and (c) having richly imagined possessing, or having lived as if one possessed, the good thing. Like maybe I could be jealous when my celebrity crush starts dating some Hollywood hunk.
And I love your point about how close envy and admiration can be!
Thanks for your reply. I’m not sure how much hangs on all of this, but I enjoyed the exchange.
Nice. I also like the example of the celebrity crush. I think a big part of our emotional lives are lived out in our imaginations in this way.
Just out of curiosity, is there a connection (actual or observable) between the sort of analysis you employ of the emotions and 1) phenomenological analyses generally either in the historical tradition known as “Phenomenology” that branched out from Husserl or in the looser contemporary sense of phenomenology or 2) Husserl’s style of phenomenological analysis in particular? I ask just because, ex. in the way you try to carve up the valences in thinking about an emotion like sadness, it sounded a bit like the sorts of analyses you might find in Husserl in trying to get at the deeper sense of our experiences.
Hi F.N.
The idea of bodily feelings as conveying affordances is quite in line with Merleau-Ponty. I also reference the phenomenological tradition when discussing expression/empathy in Chapter Five, and Scheler’s claim that we literally see the emotion in the face. However, it’s mostly just phenomenology in the looser sense of examining experience.
Hi there!
I’ve now written up a precis of my book, based on these blog posts (plus some extra bits). Hopefully this will be useful for anyone wanting to teach the view, or just get a quick overview.
You can find the precis both on Philpapers here.
and academia.edu here.