Wednesday, April 30, 2008

Rizzolatti & Craighero (2004): Class discussion summary #4

The section titled, "Evidence in Favor of the Mirror Mechanism in Action Understanding" (p. 173) gets to the heart of the matter. So what's the evidence?

The first point made by RC is that the "simplest, and most direct, way to prove that the mirror-neuron system underlies action understanding..." -- namely to lesion the relevant area(s) and show that monkeys can no longer understand actions -- is not a feasible research strategy. RC provide three reasons why this is the case:

"First, the mirror-neuron system is bilateral and includes ... large portions of the parietal and premotor cortex." (p. 173). Hmmm. It's interesting that the Parma group report that "... inactivation of a large part of F5 [on one side] produced a bilateral deficit in preshaping and grasping." (Gallese, Fadiga, Fogassi, Luppino, & Murata. (1997). A parietal-frontal circuit for hand grasping movement in the monkey: evidence from reversible inactivation experiments. In "parietal lobe contribution to orientation in 3D Space." P. Thier and H.-O, Karnath (eds). Springer-Verlag, Heidelberg. p. 264, italics theirs). If you can disrupt grasping, why wouldn't a similar deficit be evident in the understanding of grasping actions?

"Second, there are other mechanisms that may mediate action recognition..." (p. 173). Why do we need the mirror system then? See previous post.

"Third, vast lesions as those required to destroy the mirror neuron system may produce more general cognitive deficits that would render difficult the interpretation of the results." (p. 173). See my comment on their first reason.

Sounds to me like somebody needs to do the critical experiment.

So, instead of using the "simplest, most direct" method to test the MN theory, RC have to resort to correlative methods: "If mirror neurons mediate action understanding, their activity should reflect the meaning of the observed action, not its visual features." (p. 173). The problem here is the usual problem with correlation studies: there's no way to assess causality. Do MNs cause action understanding? Or is their activity just correlated with action understanding that is achieved by "other mechanisms that may mediate action recognition." So none of these studies actually test the hypothesis.

RC describe two such studies. In one it is shown that F5 neurons respond to action-associated sounds (ripping paper). This shows that sounds can be associated with actions. Cool, but it doesn't prove a thing regarding understanding. (We're reading this empirical paper for next week.)

The other is potentially interesting. Here's the logic: "If mirror neurons are involved in action understanding, they should discharge also in conditions in which [the] monkey does not see the occurring action but has sufficient clues to create a mental representation of what the experimenter does." (p. 173). The study by Umilta et al. (2001, Neuron, 32: 91-101), recorded cells in a full vision condition (monkey sees an action toward a visible object), and a hidden condition. The hidden condition was this: while the monkey is watching, the experimenter places a piece of food behind a screen. The action is then directed toward an object that the monkey can't actually see. (Recall that pantomiming actions doesn't activate MNs.) The result was that "more than half of the tested neurons" (p. 174) responded during the hidden condition. RC conclude, "It was ... the understanding of the meaning of the observed actions that determined the discharge in the hidden condition."

What can we conclude from this result? First, following the logic of the authors, the study shows that a little less than half of mirror neurons are NOT involved in action understanding. Second, we can conclude that monkeys can mentally represent objects in working memory (no surprise) and that this representation can interact with response properties of (some) MNs. There is no evidence that the MNs supported the understanding of these actions, as the actual understanding of the hidden action could have been achieved by the OTHER action understanding system.

RC conclude the section by stating that "... the activity of mirror neurons correlates with action understanding" (p. 174). I don't think action understanding was ever measured, and only about half of MNs seem to correlate, but even ignoring these limitations, correlation doesn't test the hypothesis and so proves nothing.

So, in fact, the main conclusion from the "Evidence in Favor of the Mirror Mechanism in Action Understanding" section is that there is no evidence.

3 comments:

Anonymous said...

May 2, 2008

MN radical skepticism rejects the very premise that MN exist at all. Instead, the putative discovery is treated as an artifact of the scientific instrumentation, i.e., the neuron-detecting probe. Having discovered, with this exquisite multi-channeled artifact, individual neurons which respond reliably in laboratory conditions, it is natural for the authors to conjecture imaginatively that their clever device has discovered a new type of neuron for the first time.

Unfortunately, their experiments are negated by their use of mature animals in the experimental preparations. Unlike the landmark psychology experiments with neonate kittens, deprived of movement and so failing to develop vision (proving that motorsensory feedback is necessary to learn this perceptual ability), the MN animals were mature participants in their primate center world. Of course they had imitation-based perceptual recognition (“findings” to the contrary). No animal can exist socially without it!

Is this because of a special kind of neuron, or because of a lifetime of observation and learning? Parsimony demands the latter. The so-called MNs are probably no more than markers in a complex, learned cognitive network.

The upside of the MN fad is to focus attention on the need to analyze the component of imitation in language learning. This is surely the key to motorsensory recruitment. See Iacoboni and Wilson, Cortex 2006, for example.

Remember that speech is volitional activity. That means it embodies intentionality, and so links to the entire social identity of the speaker. Of course it is ‘cultural’ in the extreme, where ‘cultural’ must include the complex socialization of primates and other mammals like elephants. This is the stuff of ‘software’, not ‘hardware’, in the brain (is learning hardwired, requiring Learning Neurons?)

The critical research among humans would be to investigate the language acquisition deficits of hearing mutes. The prediction would be, that denied motor feedback for phonological perception, their lexical semantic integration/creation would be impacted. It’s a given that they would have output difficulties. The motorsensory hypothesis prediction is that there would also be severe impairment in acquiring lexical semantic comprehension, even though hearing is spared.

---caludy

Greg Hickok said...

I don't think I would go so far as to say that MNs are an artifact of scientific instrumentation. Rather, their response properties are quite interesting and theoretically relevant.

Anonymous said...

Of course MN's properties are very interesting. It's the edifice of speculation that has grown up in their name which is is being questioned.

The point of the comment was to introduce the viewpoint of MN radical skepticism ("MN-not" for short). So far the discussion summaries seem consistent with this position. Why not go all the way?

The speculation about origins is a throw-away. It just seems obvious that if you head a very expensive lab, specializing in single-cell recordings of (human-like!) primates, you will be heavily invested in reifying the neuron.

Neurons are fine for bona fide brain processes like perception and motion, but when scientists leap to speculation about intentions (mirroring! imitating!), they have crossed the line into metaphor. These people seem to believe they have stumbled onto the Consciousness Neuron!