Social: Why Our Brains Are Wired to Connect (9 page)

Figure 3.3
Cyberball

We had people play
Cyberball
while they were inside an fMRI scanner.
The subjects believed that they and two other individuals were simultaneously having their brains scanned while they played the videogame over the Internet.
We told them we were interested in how brains coordinate with one another to perform even simple tasks like ball tossing.
The individuals had no idea they were about to get rejected in the scanner.
But after a few minutes of throwing
the ball around, the other “players” stopped throwing the ball to the actual participant.

After participants were rejected, they got out of the scanner, and they were taken to a room to answer questions about their experience.
Frequently, these individuals would spontaneously start talking to us about what had just happened to them.
They were genuinely angry or sad about what they had gone through.
This was unusual for an fMRI study back then because most tasks didn’t generate personal emotional reactions.
We had to pretend that we hadn’t been paying attention to what had happened in the scanner because we did not want their answers to the questions they were about to be asked to be contaminated by anything we might say.

We spent the better part of the next year analyzing the data, but
there was a single moment when we knew
we might be on to something interesting.
Naomi and I were in the lab late one night, and my graduate student Johanna Jarcho was analyzing her data from a physical pain study on the next computer.
We were all looking back and forth between the two data sets when we noticed a striking similarity in the results.
In the physical pain study, participants who experienced more pain distress activated the dACC more.
The same was true in the social pain study, as participants who experienced more social distress when rejected activated the dACC more.
In the physical pain study, participants who activated the right ventrolateral prefrontal cortex experienced less physical pain.
Similarly,
in the social pain study, participants who activated
the right ventrolateral prefrontal cortex experienced less social pain.
Finally, in both studies, participants who activated the prefrontal region more activated the dACC less.

Both studies were telling us the same thing.
When you experience more pain, there is more activity in the dACC.
Lots of studies had shown this before ours—but ours was the first study to show that this was true not only for physical pain but for social pain as well.
In both cases, a person’s ability to regulate the distressing aspects of pain was associated with increased ventrolateral prefrontal
activity, which in turn seem to mute the dACC response.
Looking at the screens, side by side
, without knowing which was an analysis of physical pain and which was an analysis of social pain, you wouldn’t have been able to tell the difference.

These findings highlighted one of the things fMRI research can do to help us understand the human mind in general.
It can illuminate when two mental processes that seem different actually rely on common neural mechanisms, suggesting they are more psychologically intertwined than we would have guessed.
Here, the mammalian need to recognize social threats appears to have hijacked the physical pain system to do what the pain system always does—remind us when there is a threat to one of our basic needs.

What Does the dACC Really Do?

When our
Cyberball
paper was published, it propelled our careers.
Newspapers and television shows wanted to interview us.
A number of documentaries being made about pain or social connection wanted to include a segment about our work.
We even got invited back to the conference in Australia that had inspired the study so that Naomi could present it.

Nevertheless,
lots of scientists didn’t buy our findings
that the dACC supported the experience of social pain or that social and physical pain shared underlying processes.
It’s natural for scientists to be skeptical of a finding before it is replicated.
But in our case, the skepticism was less about waiting for replications and more about not believing the story was plausible.
At the time, the dominant theory of dACC function implied that it had little to do with pain processing, social or physical.
This account largely ignored all of the cingulotomy and animal work from the 1950s as if there were a statute of limitations on the validity of those scientific findings.

In the mid- to late 1990s, several neuroimaging studies were published
suggesting that the dACC performed two closely related
cognitive functions:
conflict monitoring
and
error detection
.
Here’s a simple demonstration.
Say the following words out loud:
now, how, cow, wow, mow
.
If you hesitated when you got to
mow
but pronounced it correctly, that’s conflict monitoring (that is, you detected that there was a conflict between your impulse and the correct response).
If you pronounced it incorrectly and then said, “Oops, that’s not a word—it was the leader of the Chinese Communist revolution,” you just engaged in error detection.

In 2000, a scientist named George Bush (no relation to the former presidents) published
a seminal paper on the function of the dACC
.
Citing many neuroimaging studies of cognitive control, he too concluded that the dACC plays a key role in cognitive processes like conflict monitoring and error detection.
It was a conclusion that still holds up very well a decade later.

Bush’s review also concluded that the dACC does
not
play a role in emotional processes.
Processes related to emotion were identified with another part of the cingulate: the rostral anterior cingulate cortex (rACC).
Superficially, this seemed like a parsimonious division of labor.
Psychologists have long enjoyed dichotomizing processes
into cognitive and emotional variants (such as thinking versus feeling), as if these were mutually exclusive phenomena.
Bush drew this conclusion from several studies that appeared to show emotional processing localized to the rACC, but not to the dACC.
But that conclusion doesn’t hold up even based on the data that was available back then.
All but three of the emotion studies reviewed focused on psychiatric populations, who may not be representative of how healthy brains respond.
The majority of the nonpsychiatric studies included actually showed that the dACC
was
involved in affective processes.
Moreover,
several other neuroimaging papers of emotion or pain distress
that were left out of the review, but that had already been published at the time, clearly pointed to dACC involvement in emotion or pain distress.
As preferable as it might have been to link the dACC with cognition and the rACC with emotion, the truth is more complex.

Our Alarm System

A year after our first paper on social pain came out, Naomi and I published
a paper on a new model of dACC function
that sought to characterize both the affective and the cognitive functions of this region.
We characterized the dACC as an
alarm system
.

Let me tell you about a few of the lousy alarms in our house to illustrate what is necessary for a good one.
We live in an older home with some quirks that still have not been fixed since we moved in a few years ago.
First, we have a doorbell on our front door that doesn’t work.
If you stand close to the front door, you can hear a whisper-level sound of a circuit connecting when someone pushes the button, but that is all you will hear.
Until the pizza delivery guy realizes he should try the metal knocker on the door, he just waits, assuming we can hear the doorbell when, in fact, we can’t.
I know we should get it fixed, but
everyone figures out to use the metal knocker
, so we have never been particularly motivated to do so.
We also have a smoke detector that goes off every once in a while even when there is no smoke.
This is especially annoying when the every once in a while is at 3 a.m.

These are both terrible alarms; each is missing one of the two vital components of a functioning alarm mechanism.
An alarm needs a
detection system
that keeps track of whether some condition has been met or not.
Smoke detectors commonly use a photoelectric detector that consists of an unbroken beam of light hitting a photocell.
When a sufficient number of smoke particles break the light beam, smoke has been detected.
Given that our smoke alarm goes off at random times when there is no smoke in the house, there is something wrong with its detection system.
An alarm also needs a
sounding mechanism
that is triggered by the detection system.
The sounding mechanism in our smoke alarm works fine, obviously.
But the sounding mechanism in our doorbell doesn’t function, so we don’t know when someone is at the door.

In our neural alarm system model, we proposed that the dACC is an alarm system that serves both to detect a problem
and
to sound an alarm.
The smoke alarm needs to let everyone in the vicinity know that there might be a fire, to call 911, or just to flip the burgers so they stop burning.
It has to be able to interrupt whatever else you are doing or focusing on.
This is precisely what emotions do for each of us.
The conscious distress of physical pain motivates us to take our hand off the stove; the pain of social exclusion motivates us to work to reconnect with others.

Detecting conflicts and errors is often a source of emotional experience.
Getting a B on a test isn’t intrinsically emotional, but if you expected to get an A+, it will most likely cause distress.
It occurred to us that the conflict monitoring and error detection studies that pointed to the dACC’s role in cognition might have also produced emotional responses, but perhaps the studies overlooked them because these emotional responses were never measured.
So we decided to measure them.

Bob Spunt, then a graduate student in our lab, ran an fMRI study with Naomi and me in which he used
a conflict monitoring/error detection procedure called the
stop-signal task.
(This task is a variant of the
go/no-go task
described in
Chapter 9
.) On most trials the task was incredibly simple.
An arrow appeared on a computer screen pointing to the left or right; when it did, a corresponding key on the keyboard had to be hit as quickly as possible (one key for left, one key for right).
These trials went by at a rapid clip, about one per second, and they were easy.
A quarter of the trials, however, required a different response and were trickier.
On these, a
stopsignal tone
was played after the arrow appeared.
This tone indicated that participants should ignore the arrow and not press any button on that trial.
It was a signal indicating that the participants should stop, just for that one arrow.
This is akin to a traffic light turning yellow just as you are getting to the intersection; the changed light indicates that you need to override the plan you have already set in motion.
On early trials, the tone was played about 250 milliseconds
after the arrow appeared.
If this gave participants enough time to stop themselves from hitting an arrow key, the tone was shifted so that it came later.
The task kept changing until the tone came long enough after the arrow key that participants couldn’t help but mistakenly hit the arrow key when they shouldn’t have, about half of the time.
Participants couldn’t win.
The better they were at this, the harder the task became.
Personally, I find the task absolutely maddening, which is why it was perfect for our purposes.

After every 16 trials that included 4 of the dreaded stop trials, participants were asked to what extent the just-completed block of trials had made them feel anxious and frustrated.
There were also
go-only blocks
that included no stop trials, and participants were always informed which kind of block was coming next.
People knew whether the upcoming block was going to have the annoying stop trials or not.

In Bob’s first analysis, he demonstrated that the
error trials
(that is, when people were meant to stop but failed to) produced a strong response in the dACC, just as countless prior studies had.
Next, he used the frustration that participants expressed at the end of each block to see if there were brain regions whose activity was stronger during errors that were more frustrating, compared to errors experienced as less frustrating.
Although the task didn’t change much from block to block, people did report some blocks being more frustrating than others, and the activity in the dACC tracked this.
The more frustrating the errors, the greater the dACC activity.
No other region in the brain, besides the dACC, tracked the frustration participants experienced during the errors on this task.
We also found some evidence suggesting that even on the other trials that did not require stopping, the dACC produced greater activity to the extent that participants were anxious.
In other words, as participants became more anxious about the prospect of stop trials, we saw evidence of their anxiety in the dACC responses.