ACChey Breaky Heart

October 12, 2003

It is a common feature of many languages that emotional distress is described in terms of pain: we speak of “hurt” feelings and heart “ache”. Poets would argue that this shared language reflects our deeper shared experience as human beings. Now neuroscientists do too.

A study making news this week (Eisenberger, Lieberman, and Williams Science v302, p290) provides evidence for some common neural circuitry in the manifestation of physical and emotional pain. The focus is on the emotional distress caused by social exclusion:

[T]he notion that the pain associated with losing someone is similar to the pain experienced upon physical injury seems more metaphorical than real. Nonetheless, evidence suggests that some of the same neural machinery recruited in the experience of physical pain may also be involved in the experience of pain associated with social separation or rejection. Because of the adaptive value of mammalian social bonds, the social attachment system, which keeps young near caregivers, may have piggybacked onto the physical pain system to promote survival.
The study focuses on the pain of social snubbing, so the connection to loss or separation and the just-so evolutionary story here are compelling but not fully supported. This seems to me a nicely done piece of work, but if you read the scientific press you might take the results as stronger than (I believe) they are. It is a solid piece in the puzzle rather than a conclusive demonstration.

(The paper is not freely available on line. For more detailed news coverage, see here and here; the paper's Materials and Methods section is also accessible.)

Eisenberger et al. used a clever task to study exclusion: a virtual ball-tossing game, called “CyberBall”, in which the participant plays catch on a computer screen with two other players. The other two players were actually a computer program, but participants were told that they were other subjects being studied simultaneously. (I'm not sure what the experimenters had to say to make people accept this. It doesn't seem to me a very believable cover story.)

Within this framework, the experimenters could manipulate the participant's inclusion in the game. They chose three different conditions. In implicit exclusion, participants watched the other two players play without taking part; they were told they could not connect to the game because of technical difficulties. In inclusion, the participant played and was passed the ball frequently. In exclusion, the participant received seven tosses but was then shut out of the game as the other two players passed it only between themselves for about forty-five throws. The experiment went in order through each of these conditions.

The participants played CyberBall while being scanned in a Magnetic Resonance scanner. The study uses functional Magnetic Resonance Imaging (fMRI) to identify brain areas that are differentially active during performance of these related tasks. Physiological changes in the brain during neural activity are detectable as small differences in measured Magnetic Resonance signal at corresponding locations; analysis of the fMRI data can thus reveal which areas show more activation in one task, such as Exclusion, than in another (control) task, such as Inclusion. Such differential activation is then ascribed to a localized difference in brain function required for one task but not the other. (There are caveats in the interpretation of fMRI results which I won't go into here; for now, assume that it all works as advertised.)

Eisenberger et al. found an interesting pattern of brain activity. In comparing Exclusion and Inclusion, two areas showed up prominently. One of these is the Anterior Cingulate Cortex, or ACC, an area that has a big role in the interface between emotion and cognition. The other is a region loosely described as “right ventral prefrontal cortex”. Don't worry about the details; I'll call this second region RVPFC, for no particular reason. In the comparison between Exclusion and Inclusion, both ACC and RVPFC were more activated in Exclusion than Inclusion, and the extent of activation in these areas was correlated (ACC positively, RVPFC negatively) with degree of distress as measured by a post-experimental questionnaire.

This pattern of activity is similar to that seen in the response to physical pain, not the sensation but the emotion (affect) of pain. ACC is implicated here as the “alarm system” that detects a problem, and RVPFC plays the role of regulator of pain distress, effectively inhibiting the ACC response. This is seen in the positive correlation with distress in ACC and the negative correlation in RVPFC. The story being told, mostly by the press but not completely, is that at some essential level physical and emotional pain (of the snubbing variety) are the same.

The main difficulty with this story is that ACC has been shown to have diverse functions. It has been implicated in emotional self-control, conflict monitoring, error detection, attentional control, focused problem solving, and adapting responses to current goals and conditions. For example, ACC activates in the conflict trials of the famous Stroop task, in which one is sees a color word ('red','blue') printed in a different color ('green', 'brown') and is asked to name the color of the ink (or pixels). Animal experiments have also shown that disturbing activity in ACC reduces a mother's effort to keep offspring nearby and reduces separation distress (measured by outcries) in offspring.

The correlation with reported distress in ACC could represent increasing attention to the conflict between expectation (preparation to receive the ball) and reality (not receiving it) instead of an increased affective response to social pain. Eisenberger et al. argue against an attentional effect on two grounds: 1. the correlation between ACC and reported distress, 2. the greater activation of ACC in the Implicit than in the Inclusion condition. These are good arguments, but I'm not quite convinced. First, distress could be an effect rather than a cause, increasing with greater awareness of conflict between expectations and reality. A model based solely on attention to conflict would show an increased activation over the course of the Exclusion condition. Second, they're arguing that greater attention is required in the Inclusion condition because you must follow the ball, but participants may give the same attention in the Implicit condition because they expect the computer connection to be reestablished at any moment. Plus, they could be attending to the emotional conflict between wanting to participate and not being able to. The experiment's cover story did not enable the experimenters to ask about distress after the Inclusion condition.

RVPFC activation does not show up in the Implicit condition, which raises the question under their model of why Implicit showed ACC activation. If a portion of the activation in ACC were associated with attention to conflict and a portion associated with distress, these data would make sense in their terms, provided that there was more ACC activation in Exclusion than in Implicit. Directly comparing the Implicit and Exclusion conditions would have been informative here, but they did not do this. (Possibly a better control then either Implicit or Explicit would be to compare trials during Inclusion when the participant was not passed the ball to the exclusion trials.) Moreover, the negative correlation between activation in ACC and RVPFC could result from greater inhibition of the latter on the former, as in Eisenberger et al. claim, but there is no direct evidence of this. The studies they site to associate RVPFC with pain distress regulation do not give sufficient spatial resolution to accept this role unambiguously.

Neuroscientific work on emotion is accumulating rapidly, and a picture of the neural basis of emotion is beginning to form. Eisenberger et al. offer an interesting piece of this picture. Their conclusions are reasonable and fit the data, but given the diverse roles played by the ACC and the imprecision with which their RVPFC region has been characterized, I'd like to see them rule out the alternatives more definitively. Here's a thought experiment that might show a similar pattern of activation without any social pain. Call it the Rosencrantz and Guildenstern experiment. A subject flips a fair coin repeatedly, but after some number of flips the coin begins to come up heads and continues to do so. If it could somehow be arranged for subjects to believe that the coin is fair, I would predict a similar pattern of activation with increasing attention to conflict between expectation and reality (and concomitant “distress”) as the run of heads continues. (A realistic analogous task would be required in practice.) If this experiment showed the same pattern, then we could still say that the response to physical and emotional pain share a common machinery, but since that machinery would be common to other processes as well, the interpretation would not be as satisfying.