Saturday, October 22, 2011

Activation of the Hate Circuit While Reading 'Depression Uncouples Brain Hate Circuit'

A recent article published in Molecular Psychiatry has the curious title, 'Depression uncouples brain hate circuit' (Tao et al., 2011). Hate circuit, you ask? Is there really any such thing? Is the existence of a distinctive brain circuit for hate so well-established that we ought to go about including it in the title of our papers? And what does it mean for this circuit to be uncoupled in depression? That depressed people no longer have coherent feelings of hatred?

The current article refers to the one prior fMRI study on the topic, which examined the 'Neural correlates of hate' (Zeki & Romaya, 2008). The 17 participants were chosen because they expressed intense hatred for a particular individual. Sixteen people hated an ex-lover or a competitor at work, and one person hated a famous political figure (see Hate On Halloween for details of that study). Participants viewed pictures of a person they hate, and the resultant BOLD signal changes were compared to when they viewed pictures of neutral people.

And the groundbreaking hypothesis? Love and hate might be represented by different brain states! Who knew?
We hypothesized that the pattern of activity generated by viewing the face of a hated person would be quite distinct from that produced by viewing the face of a lover.
The results identified 7 regions that were significantly more active for the Hated Face condition than for the Neutral Face condition. The flaming figure above illustrates a few of them, including the medial frontal gyrus [the anterior cingulate cortex (ACC) and the pre-SMA], the right putamen, and bilateral premotor cortex. The other regions were the frontal pole and our friend, bilateral insula [activated in all sorts of conditions from speech to working memory to reasoning to pain to disgust to the allure of Chanel No. 5 and "love" of iPhones]. An additional correlation analysis related degree of hatred to level of activation across 5,225 voxels (using an uncorrected statistical threshold of p≤0.01) and found three regions to be most related: right insula, right premotor cortex, and right ACC.

These results set the stage for the current study by Tao et al. (2011), which compared the resting state functional connectivity patterns between controls and severely depressed individuals. The "resting state" or "default mode network" (DMN) is the brain activity observed when there is no active task (Raichle et al., 2001). In other words, the participants are free to daydream about their lover or to think about dinner or to remember the amusing movie from last night or to focus on feelings of despair. A specific group of brain regions has been identified as the DMN, and these are deactivated when participants have to perform a demanding cognitive or perceptual task.

A new feature of the present study was the community mining algorithm used to determine coherent resting state networks among the 90 regions of interest (ROIs). First, a template was formed based on data from 37 healthy controls. Then the network connectivity for the control template was compared to that of two depressed groups: 15 unmedicated first-episode major depressive disorder (FEMDD) patients and 24 resistant major depressive disorder (RMDD) patients.

The 6 "communities" or resting state networks are illustrated below. Note that RS1/DMN (red in the top figure) isn't identical to the typical DMN (orange in the bottom figure).

Top: Adapted from Fig 1C (Tao et al., 2011). Left: Medial view of the surface of the brain. Right: The lateral view of the surface of the brain. Different colors represent different communities. Bottom: Adapted from Buckner et al. (2008).

One big difference between the two schemes is that the dorsal ACC /SMA active task network (blue in the bottom left figure) is part of the DMN in Tao et al.'s community structure (red in the top left figure).

But wait, where is the 'hate circuit'?? It emerged with some bizarre post hoc hand waving:
It can be seen from Figures 2, 3 and 4 that the strongest evidence for reduced connectivity compared with control subjects in both FEMDD and RMDD is that between the insula and putamen in both brain hemispheres (s=0.4 and 0.25 for FEMDD and RMDD, respectively). Additionally, the link between the left superior frontal gyrus and the right insula is also reduced (s=0.2991 and 0.2658). Thus, the links between the three main components of the ‘hate circuit’ have become largely uncoupled.
OK, I thought Zeki's 'hate circuit' included bilateral premotor cortex and the frontal pole, plus the putamen only in the right hemisphere. But somehow, the community mining algorithm determined that the insula [part of RSN4 - auditory network] has links with the putamen [RSN6 - subcortical network] and the dorsal superior frontal gyrus [RSN1 - DMN, and perhaps not the same area as in Zeki & Romaya] only in controls but not in the depressed participants. And this set of links in controls comprises the 'hate circuit' and nothing else.

Adapted from Fig 4a (Tao et al., 2011). The common links of the first-episode major depressive disorder (FEMDD) and resistant major depressive disorder (RMDD) networks. Red lines are links that appear in depression network only while blue lines are links that appear in normal template only. The widths of the lines are proportional to the scores.

I was critical of the original 'neural correlates of hate' study 3 yrs ago, but at least the authors never used the term 'hate circuit' in their paper1. Plus, they had something interesting to say about the overlap between brain regions engaged by looking at pictures of those you love and hate:
Our studies did indeed reveal a basic pattern. As far as we can determine, it is unique to the sentiment of hate even though individual sites within it have been shown to be active in other conditions that are related to hate. The network has components that have been considered to be important in (a) generating aggressive behavior and (b) translating this behavior into motor action through motor planning. Finally, and most intriguingly, the network involves regions of the putamen and the insula that are almost identical to the ones activated by passionate, romantic, love.
Call me a hater, but maybe the 'love circuit' has come undone in the depressive folks instead... But really, if a clinical group is lacking functional coupling between the insula and the putamen, it can mean any number of things, not only that their so-called 'hate circuit' is incoherent.2 Another important question emerges from the control data: what does it mean to have strong functional connectivity within the 'hate circuit' at rest? That you spend a lot of free time filled with hate? One of the authors, Professor Jianfeng Feng, had this to say:
MRI study finds that depression uncouples brain’s hate circuit

. . .

“The results are clear but at first sight are puzzling as we know that depression is often characterized by intense self loathing and there is no obvious indication that depressives are less prone to hate others. One possibility is that the uncoupling of this hate circuit could be associated with impaired ability to control and learn from social or other situations which provoke feelings of hate towards self or others. This in turn could lead to an inability to deal appropriately with feelings of hate and an increased likelihood of both uncontrolled self-loathing and withdrawal from social interactions. It may be that this is a neurological indication that is more normal to have occasion to hate others rather than hate ourselves.”
Now it is a seriously long stretch to conclude, from the present resting state fMRI data, that the depressed groups are impaired in controlling or learning from hate-filled social interactions, and as a consequence transform these disjointing feelings into self-loathing. Perhaps it will be more instructive to read about 'Demoralization: its phenomenology and importance' instead...


1 That phrase was reserved for the university press release, although it's unclear who was actually responsible for coining it.

2 While we're at it, why are the left inferior frontal gyrus (opercular), left precentral gyrus , right inferior frontal gyrus (opercular), and right supramarginal gyrus considered the "risk/action" circuit?


Buckner RL, Andrews-Hanna JR, Schacter DL. (2008). The brain's default network: anatomy, function, and relevance to disease. Ann NY Acad Sci. 1124:1-38.

Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman GL. (2001). A default mode of brain function. Proc. Natl. Acad. Sci. 98: 676–682.

Tao, H., Guo, S., Ge, T., Kendrick, K., Xue, Z., Liu, Z., & Feng, J. (2011). Depression uncouples brain hate circuit. Molecular Psychiatry DOI: 10.1038/mp.2011.127

Zeki S, Romaya JP. (2008). Neural correlates of hate. PLoS One 3(10):e3556.

Subscribe to Post Comments [Atom]


At October 23, 2011 1:44 AM, Blogger Neuroskeptic said...

The overlap of the love and hate circuits has been described by eminent neuroscientist Jay Z, in Z et al

I quote
"I'm running New York, I go the mayor on my pager,
You can’t fade us, you haters,
I need you, stay back, I breathe you,
Like air-ah, air-ah, air-ah,
Yeah yeah yeah yeah-a
Yeah yeah yeah yeah-a
Where are my haters, I love all my haters
Love all my haters, I love all my haters,
Uh uh uh, haters,
hater hater..."

At October 23, 2011 4:33 AM, Anonymous Anonymous said...

My hate circuit is running amok after reading about this piece of blobology nonsense. Thanks for the post.

At October 23, 2011 6:07 AM, Anonymous Anonymous said...

Neuroscientists are idiots.

At October 23, 2011 12:01 PM, Blogger The Neurocritic said...

Those who make sweeping generalizations about neuroscientists are idiots.

At October 24, 2011 2:28 AM, Blogger Finn Årup Nielsen said...

The Neurocritic writes: "An additional correlation analysis related degree of hatred to level of activation across 5,225 voxels (using an uncorrected statistical threshold of p≤0.01)..."

To some defense of Zeki & Romaya one would note that the t-values are somewhat high for the correlation analysis. The highest is t=6.37. That brings the p-value to 0.023 when viewed wrt. the search volume (corrected for that "small" volume I suppose? Shouldn't we have a ban on small volume correction?). The right panel of Figure 5 seems, however, to be a voodoo correlations scatter plot...

At October 24, 2011 10:12 PM, Blogger Scott Kolbe said...

I thought Molecular Psychiatry was a reputable journal. As a neuroimaging researcher it is offensive to see the technique bastardised this way by researchers who obviously have no idea how to interpret their data. The editors should have pinged this one out of the gates and one would have to question whether the reviewers were impartial and/or expert.

At October 25, 2011 6:02 AM, Blogger Marsha Lucas, PhD said...

I would have been done with my dissertation in far less time if only I'd mastered the fine art of "bizarre post hoc handwaving."

At October 27, 2011 10:14 AM, Anonymous Anonymous said...

Perhaps there is a simple explanation: the title was written by the Chinese authors...

At October 27, 2011 10:18 AM, Anonymous Anonymous said...

Folks, as more and more of this nonsense accumulates the reputation of our discipline sinks. Lack of replicability, sloppy conceptualization, amateurish analyses by people with no background in neuroscience... They all contribute to slowly corrode the status of our field (which indirectly will have an impact on funding)... I'm taking on more papers to review than I would like to, simply because otherwise they may end up being reviewed by some pinheads with no clue about this stuff...

At February 01, 2018 2:36 PM, Blogger Unknown said...

Seeing as most health evaluations are supposed to stem from a stress-vulnerability model these days, it would make sense to say that if you want to create an emotive brain compass, go for more stressed and less stressed. So, dmn is less stress, and any other activation whatsoever is stress. It would probably be the case also that inertia is stored as trauma, with more stress affected regions requiring more amgydala retraining by the brain, eg. through CBT, sleeping, meditation, cessation of smoking, etc. As long as the brain can deal with it's inertia with minimal work to do, the better it can function when it needs to- to survive, when it is pressed to function at all, eg. Growing dopamine receptors "in order to prompt searching for food" after fasting for a few hours. If it is functional to be one way over another, it helps survival.


Post a Comment

Links to this post:

Create a Link

<< Home

eXTReMe Tracker