A prototypical “locationist” theory of emotion. Amygdala/fear (yellow), Insula/disgust (green), OFC/anger (rust), ACC/sadness (blue). Credit: Lindquist et al/Cambridge University Press.
In the past I’ve written about a major problem with the brain’s fear “center”: specifically, fear can be experienced even when this center (the amygdala) is destroyed! Clearly then we need to re-think what we mean when we talk about brain “centers”.
Take, for example, the fact that there are brain regions that encode specific senses–sight, sound, touch, smell and taste are all encoded in distinct regions of the brain’s neocortex. There are also specific regions of the brain for “motor” functions like moving the body’s muscles. But, many people would be surprised to learn that we haven’t directly mapped out psychological functions to specific brain regions. In fact, just because a brain region is required for a function of the mind, doesn’t mean that that region is entirely responsible for causing this same mental function .
We can explain this counter-intuitive idea by recognizing that even if a piece of information is encoded in a region of the brain (like visual information in the visual cortex), this information may not be accessible to the rest of the brain or might be selectively accessible to only certain regions. For example, while your eyes may “see” most of the information in front of you, only a very small part ever makes it to your conscious awareness. Most of what we “see” is filtered out by selective attention. On the other hand, some of this same unattended information may be passed along to brain regions which process information without your conscious awareness, including brain regions like the amygdala (the so-called “fear center”) which may act to focus conscious attention on important things in the environment (e.g. food, snakes, shiny new toys).
However, when it comes to the experience of either simple or complex psychological states, much broader regions of the brain are almost certainly involved. In order to fully experience something, we almost certainly need to be conscious of it.
Searching for mind in the brain
Let’s take the example of emotion.
Last year, cognitive neuroscientists conducted a comprehensive meta-analysis of neuroimaging studies of emotion. They included research looking at the neural correlates of five key “basic” emotions: happiness, sadness, fear, disgust and anger. What they found was striking: many of the same brain regions were activated across multiple emotions! In particular, the amygdala appears to be metabolically active in tasks that evoke various distinct emotions.
Based upon this data, they criticize both a) “locationist” theories of emotion which propose that specific psychological functions are “located” in specific places (e.g. fear “centers”) as well as b) the idea that emotions are discrete categories (e.g. that fear feels distinctly different from anger which feels different from sadness, etc). Instead they argue that emotions don’t rely upon one specific brain region but many (point a, which I allude to above), but they also take a further position to argue that these brain regions overlap so much that the brain cannot tell apart the “basic” emotions we all know so well. Instead they believe there are simpler psychological ingredients that can be used to create the emotions we label “happiness” and “fear”.
I will address this latter idea first, because it’s pretty controversial, whereas the “locationist” approach to emotion is actually fairly uncommon among neuroscientists.
Every flower in this image can be divided into tiny points of light, representing a single dot of red or a single edge along the curve of the flower. Credit: Eistreter (Own work), via Wikimedia Commons
We should start by recognizing that the mind is almost certainly composed of simpler ingredients than the experiences that we can label. If the mind relies upon processes of the brain, then brain events can obviously be described at levels of such intricate detail (e.g. ions flowing across membranes) that we could not possibly “experience” them. At a slightly more complex level though, let’s consider an analogy to the visual system. For example, it is believed that visual images are composed of simple features like colors (e.g. red, green) that are extracted from the visual spectrum of light or of more concrete features like the “edge” of a picture frame. Generally we perceive the edge of a picture as a 3-dimensional line or corner, rather than as points of light reaching orientation-responsive neurons in the visual system of the brain. But if we try to imagine every single point of light along the edge of the frame as activating one or more neurons in the visual system, it’s almost conceivable to imagine that we might be able to break down that experience the tiniest bit of edge. Similarly, I can also imagine dipping a paintbrush in a big blotch of bright red paint and then watching the smallest perceptible molecule of red slowly trickle from its tip and fall towards the ground.
In a sense then, we can probably imagine that our typical experiences might be broken down into simpler, psychologically identifiable pieces at some level.
How would this work for emotions?
Facial Expression of anger. Credit: Gianmaria Zanotti (Reportergimmi), Via Flikr
Generally, the most popular theory of emotion breaks emotions down into five or six “basic” emotions: fear, anger, sadness, happiness, disgust and surprise (the last is sometimes considered to be a component of other emotions). The supporting evidence is fairly diverse, but includes stereotypical facial expressions of emotion and stereotypical physiological responses, a practiced ability to identify emotional expressions on other people’s faces (although, there is convincing evidence that extreme emotion cannot be identified by facial expression alone; think the grimace of an athlete after winning a big match), and there tends to be a fair amount of consistency across cultures and early-life development. In other words, even if these prototypical emotion categories are artificially constructed by people, there’s reasonable evidence that these categories do a pretty good job.
In the present paper, the authors argue that these categories can be broken down quite a lot further.
Facial Expression of fear. Guillaume Duchenne, via Wikimedia Commons
Just like we can break down a picture into its parts or a blotch of red into a tiny droplet, it might be possible to break down emotional states into individual pieces (e.g. muscle tension, heart racing, brow pulsing and sweating, skin prickled with goose-flesh… these probably all contribute to an emotion like anger). While they aren’t clear on the best way to define these ingredients, the authors settle on what they call “valence” (positive or negative) and “core affect”, which is a measure of some combination of all the sensory cues we use to figure out how we feel (e.g. internal feelings, how the body and muscles feel, heart racing/pumping slowly, etc). They further argue that these internal feelings or only become emotion when they are assigned some kind of label by the brain’s knowledge system, a process called “conceptualization” or “categorization”. The authors call this the “psychological constructionist” approach, because emotions are being “constructed” out of smaller pieces.
This proposal actually isn’t as strange as it sounds. Take, for example, some of the more complex emotions we have that appear to overlap. How do we distinguish anxiety from excitement? Very similar internal feelings may be evoked by both states (e.g. butterflies, muscle weakness, skin tingling). In fact, these two forms of anticipation can even be inter-changed to some extent: many of us will be familiar with the experience of trying to change anxiety into excitement by altering how we visualize an upcoming event like a presentation or interview (“Picture yourself being successful!”).
But of course the question addressed in this paper is whether or not the brain represents these states differently. The authors find that many regions in the brain (amygdala, insula, orbitofrontal cortex, anterior cingulate cortex) are each activated by multiple emotions and that individual emotions also activate many different brain regions. In other words, brain activation studies suggest that various emotions activate large parts of the brain, and that brain regions are not very specific to any one emotion.
The authors argue that:
“Our meta-analytic findings were inconsistent with a locationist hypothesis of amygdala function but were more consistent with the psychological constructionist hypothesis. Our density analyses revealed that… voxels within both amygdalae… were not functionally specific for instances of perceiving fear… Our meta-analytic findings were inconsistent with the locationist account that the anterior insula is the brain seat of disgust but were more consistent with the psychological constructionist account that insula activity is correlated with interoception and the awareness of affective feelings.”
The basic idea is that because many “emotion-specific” brain regions seem to be involved in many different emotions, that the role of each region is to produce a simpler psychological ingredient. Ironically, this is is very close to a locationist approach as well. By stating that psychological ingredients are located in specific brain regions, the authors are essentially supporting a locationist paradigm, which is really a contradiction in terms.
So what can we say about this study?
One key point is that the authors are addressing a straw man. Just because locationist models are flawed, does not mean that “constructionist” models of emotion (e.g. no true emotion categories like fear or anger) are correct. In reality, the brain must construct all psychological experiences from smaller pieces, but exactly what those pieces are and whether or not they tend to cluster into categories is equivocal based upon this study.
On the other hand, does this study put the nail-in-the-coffin of brain-emotion locations? Well, I certainly think it’s a tip in the cap. The alternate to locationist theories of mental processes is to assume that the mind is “distributed”, that specific mental events occur as the interaction between distant brain regions. Such an account does not allow for any single brain structure to produce an entire feature of experience. It’s also broadly consistent with studies of patients with brain damage, generally one of the strongest pieces of support for locationist perspectives. Even if damage is specific to one region (e.g. amygdala) and only a single psychological function appears to be affected (which is very unlikely to happen in real patients), this is only evidence that a specific brain regions is critical or necessary to produce certain psychological states, but not whether this region alone is sufficient. In fact, hard locationist theories of psychological functions are pretty widely discredited at this point.
Further interpretations about discrete or “constructed” emotions are pretty hard to make based upon this study. The major problem with meta-analyses of fMRI data is the very low spatial resolution–we cannot see individual neurons, instead we can only make out big blobs. These blobs may contain thousands of neurons, some of which may be involved in producing the emotion of “happiness”, others perhaps “sadness”, etc. This could easily explain why the same brain regions are activated by different emotions, thus conflating the authors’ “construction” approach with more traditional “category” approaches. So regardless of which account is true, this result doesn’t allow us to draw any strong conclusions.
Perhaps we’ll just have to wait for that BRAIN project to solve the mystery?
