Tuesday, April 13, 2010

Recognizing facial expressions without the capacity to produce them -- Moebius Syndrome

Former TB West grad student, now Rotman Institute faculty member Brad Buchsbaum, pointed me to this interesting NY Times article on Moebius Syndrome, a congenital disorder that causes facial paralysis. Much of the article focuses on the social impact of the inability to express emotions on the face. Of particular interest here however, is the reference in the article to a new study by Kathleen Rivas Bogart and David Matsumoto on the recognition of emotional facial expression by people with Moebius syndrome. These authors report no difference between individuals with Moebius syndrome and controls in the ability to recognize emotional facial expressions. This reinforces what we've been saying here for a long time now, that in contrast to the central claim of mirror neuron theorists, you don't need to be capable of generating an action to recognize/understand that action in others.

13 comments:

sergio said...

Dear Hickok,

I do not think that this is a correct example of dissociation between production and recognition of facial expressions. As far as I remember, Moebious syndrome is due to agenesis of the motor nuclei of facial abducens nerve. In other words, it is a disease of the second (lower) motorneuron.

Anonymous said...

As wittgenstein said, better not to talk about what one does not know....

Tacchini said...

But do they have high-level facial expression "goals" despite the facial paralysis?

i.e. do they have "inner" facial expressions that are planned but never executed due to the lower level motor impairment?

Greg Hickok said...

Ah, the old, "that's not the relevant level of the motor system" argument.

First of all, had it been found that people with Moebious syndrome failed to recognize facial expressions, the mirror neuron folks would be all over it, claiming that this is a clear example of motor system involvement in action recognition. It is, therefore, worth pointing out such negative results.

Second, peripheral disruption has effects in cortex. For example, peripheral blindness affects the functional organization of visual cortex, amputation of a limb affects the functional organization of somatosensory cortex in humans, and sectioning motor nerves in animals affects the organization of motor cortex. Thus, even though the disease in Moebius affects the peripheral motor system, one expects higher-level effects.

More generally, it is worth noting that motor theories of perception have a tendency to start out fairly strong, implicating relatively low levels of the motor system and then backing away as the empirical facts roll in. This happened to the motor theory of speech perception which initially implicated the actual motor programs for generating articulatory gestures. When it was discovered that you didn't need the ability to articulate speech to perceive speech sounds, they backed down, saying it was the "intended gestures" that represented the critical level of the motor system. As I blogged about previously, the mirror neuron folks are moving in the same direction. The problem with this is that you end up backing yourself right out of the motor system.

sergio said...

I read with great interest your comments.
What you say is right. That is, in addition to the deficit of the lower motor neuron, a deficit in the upper motor neuron may coexist. I remember to have read something about it, but I do not remember what actually causes the deficit of the upper motor neuron. In other words, if the deficit of the upper motor neuron depends on the deficit of the neuron located downstream (lower motor neuron) .
Let’s accept that this is true. But does this process continue even after? In other words, if the deficit of the lower motor neuron causes a deficit in the upper motor neuron, the deficit of the latter also affects the neuron that precedes it, and so on? But in this way, to which extent will the brain “compromise” in its entirety?
Now going back to Möbius Syndrome, may the agenesis of the motor nuclei of the facial and abducens nerve be responsible for the deficit of central mechanisms underlying the production of expressive aspects of emotions? Another possibility is that Möbius Syndrome is in fact a multisystemic disease and interests central mechanisms underlying the production of expressive aspects of emotions, regardless of the agenesis of the motor nuclei of the facial and abducens nerve.

Sincerely

Sergio

Greg Hickok said...

That's a good question Sergio and I don't know what the answer is...

More generally, this discussion illustrates a fundamental problem with the mirror neuron/action understanding position. Let's think about the possible outcomes of the Moebius syndrome facial expression recognition study and their interpretation by mirror neuron theorists:

1. Facial expression is impaired.

Interpretation: MN theory of action understanding is correct.

2. Facial expression is unimpaired, neural disruption of expression execution is restricted to peripheral nervous system.

Interpretation: motor impairment does not affect the relevant part of the motor system so MN theory is still correct.

3. Facial expression is unimpaired, neural disruption of expression execution involves higher level motor system.

Interpretation A: The "motor goals" are coded somewhere else so the MN theory is still correct.

Interpretation B: The relevant MN system is impaired in Moebius, recognition is preserved due to non-mirror mechanisms, but "true understanding" i.e., "from the inside" is impaired (but not detected behaviorally) so MN theory is still correct.

No matter the result of ANY experiment, the MN theory will still be assumed to be valid. This renders the hypothesis scientifically vacuous.

brockf said...

Hi Greg,

What does need to be shown for the mirror neuron of action understanding (or intention understanding?) to be disproved? I'm a supporter for mirror neurons in both the evolution and perception of language but this question is one that has bothered me, for the reasons you point out.

Is it the case that the mirror neuron debate will only subside when one generous/clinically insane person lets researchers measure his/her individual neuronal activity? Or maybe we need two people? One who can recognize facial expressions and one who cannot?

While my knowledge of mirror neurons is admittedly weak compared to yours, my instinct would be that empirical evidence isn't going to come in the form of experiments such as this Moebius syndrome observation. With something as "low-level" and broad as mirror neurons, the answers may lie below our current level of observation.

Greg Hickok said...

It has been disproven already. You don't need a motor system to perceive speech, action understanding and action execution dissociate, and humans and animals can understand actions they cannot produce. Recording from single cells won't help because, as in the monkey, the debate is not whether mirror neurons exists, it is what they are doing. I think mirror neurons do exist in humans; I'd be surprised if they didn't. But I don't believe that they are functioning to support action understanding in humans or monkeys.

brockf said...

I guess that leaves the question: Why do mirror neurons exist?

Also, is it at all possible that the mirror system was previously the primary mechanism of speech perception but, as this ability became specialized through evolution, is now only implicated in specific, ambiguous cases such as coarticulation? Gow & Segawa (2009)'s paper on the causal relationship between motor activity and speech perception suggest this, if I remember it correctly.

Greg Hickok said...

They exist for the same reason object-oriented sensory-motor cells exist: to inform action selection and execution. Lot's of types of information is relevant to action execution, the shape and location of a cup, a verbal command (duck!), and the movements of conspecifics. There's no magic here. Everyone needs to go study the behavior of "canonical neurons" in F5 and think about mirror neurons in exactly the same way. It's just a plane old sensory-motor integration circuit.

brockf said...

Interesting. Thanks for taking the time to respond to my comments. The only challenge to mirror neurons' role as simple informants of action selection and execution would be their specific link to intention and not the actual motor movements, wouldn't it? i.e., The fact that a hand grasping for "eating" correlates with different neural activity than grasping for "placing". Likewise, multiple graspings with the same intentions (e.g., eating) correlate with identical neural activity, even if the monkey couldn't see the hand in one of the cases. This is from Umilta et al. (2001) and Fogassi et al. (2005).

I interpret this as transcending simple sensory-motor associations but I speak of these associations with a negligible background in neuroscience. May be getting carried away :)

Greg Hickok said...

Hi brockf,

It appears that the object-oriented "canonical" neurons have the same abstract motor properties, so this is not anything specific to actions.

Lincoln said...

Bit of an old post, but I'll add my 2c nonetheless.
To me (and others) it seems that mirror neurons serve for to aide perception (and of course action selection). Specifically, it seems like knowledge embodied in mirror neurons (or rather, the observer’s action system) is used to predict the actions of conspecifics. As a simple example, if you’re trying to synchronise an action with another agent (e.g., press a button in time with them waving their arm), you can’t merely rely on incoming sensory information in order to make your decision about when to execute your action. If you did this, your action execution would be hopelessly out of time. On this account, what mirror neurons provide is a means to predict the action of the observed agent so that you can select and execute your action in time. This prediction is done of the basis of the incoming trajectory as well as prior knowledge of the goal derived from some process outside of the mirror system. The mirror system uses goals, it doesn’t compute them.

If you take the example of different patterns of MNS activity in response to scenes depicting grasping in the context of “washing up” or “drinking tea”, this account can easily explain those results in terms of a different prior goals resulting in different predicted actions. Washing up goals and drinking tea goals lead to different future actions. This is what the differences in MNS activity mean. They don’t mean that mirror neurons are detecting the goals. They’re just using the goals to predict different future actions.

http://www.maccs.mq.edu.au/news/conferences/2009/ASCS2009/pdfs/Colling_thompson.pdf reference for an experiment of mine looking at action synchronisation. It contains some references for an alternative account of mirror neuron function.