Friday, April 27, 2007

In love with simplistic brain theories

The number two most-emailed article at the New York Times today is Tuesday's article, "If you want to know if Spot loves you so, it's in his tail." The article covers a study published last month in Current Biology by A. Quaranta, M. Siniscalchi, and G. Vallortigara, of the University of Bari and the University of Trieste.

Quaranta et al. studied 30 dogs by presenting them with four kinds of stimuli under experimental settings: the dog's owner, a human stranger, a "dominant" unfamiliar dog, and a cat. They found that when the dogs saw their owners, there was a statistically significant (p < .00001) bias in the direction of tail wagging; there was more wagging to the right. Seeing a human stranger resulted in a similar bias to the right (p = .00007), but with less amplitude. With the cat, the bias persisted (p = .03) but will still less amplitude. When the strange, dominant dog - a Belgian Shepherd Malinois - appeared, the dogs were more likely to wag their tails to the left (p = .014).

The authors suggest that this result ties in to one of the many theories of emotion and hemispheric asymmetry that have been proposed: one by Richard Davidson of the University of Wisconsin which suggests that "approach" and "withdrawal" behaviors are associated with the left and right cerebral hemispheres, respectively. Given how cerebrum and body are wired up, movements on the left side of the body are associated with neural events in the right hemisphere, and vice versa.

I was intrigued by this paper and Davidson's hypothesis, and was a little bit skeptical that the approach-withdrawal theory is as widely accepted as the NY Times would have it. As Sandra Blakeslee, the author of the Times piece puts it:
Research has shown that in most animals, including birds, fish and frogs, the left brain specializes in behaviors involving what the scientists call approach and energy enrichment. In humans, that means the left brain is associated with positive feelings, like love, a sense of attachment, a feeling of safety and calm. It is also associated with physiological markers, like a slow heart rate.

At a fundamental level, the right brain specializes in behaviors involving withdrawal and energy expenditure. In humans, these behaviors, like fleeing, are associated with feelings like fear and depression. Physiological signals include a rapid heart rate and the shutdown of the digestive system.
For readers who were in school in the '80s, you might remember some of the "I'm left brained, you're right brained" pseudoscience that educators used to disseminate to unsuspecting fifth graders as solid neuroscientific fact. (Alison Gopnik, writing yesterday on Slate, compares today's obsession with mirror neurons with the asymmetry craze of yesteryear.) So let's consider this "fundamental" brain asymmetry a moment:

The approach-withdrawal asymmetry hypothesis is one of several inter-related hypotheses that have been put forth over the years concerning possible asymmetric processing of emotion in the brain. (See Demaree et al., 2005 for a review.)
  1. Right hemisphere. It has long been observed that neurological patients with damage to the right hemisphere are likely to suffer affective complaints than are patients with left-hemisphere damage. Earlier theories had suggested a stronger role for the right hemisphere in all emotional processing (which may still hold up when it comes to the perception of emotional states in others).
  2. Valence model. Another class of theories argued that "positive emotions" entailed a greater left hemisphere component, and "negative emotions" a greater right hemisphere component.
  3. Approach-withdrawal. Davidson and colleagues extended the valence model by incorporated a strong prefrontal component and emphasizing behavior (or the planning of behavior) rather than simply the emotional state.
  4. Behavioral activation and behavioral inhibition. Demaree et al. also discuss a related extension to which J. A. Gray is a major contributor. This is similar to approach-withdrawal, but with some subtle differences in prediction, e.g., the left hemisphere, and not the right, would be involved in the "active avoidance" of a negative stimulus (i.e., big Belgian shepherds).
The evidence for all of these theories is equivocal, even when considering the neuropsychological, sodium amytal, and EEG data on which they were primarily based, not to mention the later brain imaging evidence that posed more serious problems.

More generally, Davidson's approach-withdrawal asymmetry hypothesis is specific to the realm of emotion and emotion-driven behavior, and specific to particular areas of the brain, most notably regions within the prefrontal cortex and the amygdalae. It does not argue for anything so fundamental about the entire "left brain" and the "right brain." Even if it proved to be an adequate description for this specific problem space, it would still be one particular asymmetry that results from one network of brain regions coming together to solve one class of problems. "At a fundamental level," brain function and behavior represent myriad neural networks and problem spaces, each of which has its own peculiar pattern of hemispheric asymmetry.

Recommended reading:
McManus, C. (2002). Right hand left hand: The origins of asymmetry in brains, bodies, atoms, and cultures. Cambridge, MA: Harvard University Press.

For some other bloggers' takes on the story:
Frontal Cortex
John Hawks


Chris Chatham said...

Very interesting blog!

I am glad you didn't throw the baby out with the bathwater when it comes to functional hemispheric asymmetry. It is true that hemispheric asymmetry theories have been simplistic in the past, but this is IMO no reason to think that there is not, at bottom, a fundamental computational difference between the types of processing carried out by the two hemispheres, particularly in the frontal cortex (where asymmetries are most likely to manifest, given that region's distance from sensory cortex). Unless you subscribe to a highly modular view of the frontal cortex, it seems reasonable to suppose there's a general computational difference between left and right regions... do you really disagree with that?

Neurozone said...
This comment has been removed by the author.
Neurozone said...

I'm definitely open to the idea that a fundamental computational difference could exist. One that I am sympathetic to is that proposed in The Two Sides of Perception. It's a more compelling hypothesis: a basic computational property (i.e., the spatial or temporal span over which information is integrated) that plays out differently in different domains and explains a variety of phenomena. You can test the predictions of this much more readily than you can something like "positivity".

That said, I don't think that we should simply assume that there is just one fundamental computational difference, as tempting as the parsimony of these ideas may be. And I definitely think we can do better than left:good::right:bad. Hopefully that doesn’t make me Jerry Fodor. :)

Thanks for reading!