Tuesday, September 22, 2009

Mirror neurons and the study of science

Tony Barnhart was kind enough to invite me to a psychology seminar yesterday afternoon that was a discussion of mirror neurons, at least partly inspired by (or inflamed by) Marco Iacoboni's August 27 talk which I attended and summarized.

I found the discussion particularly interesting in light of my current studies, as it touched repeatedly on issues of what's appropriate in science--what does and does not conform to the norms of good science.

The discussion leaders began with quotes from V.S. Ramachandran and Marco Iacoboni:
"The discovery of mirror neurons in the frontal lobes of monkeys, and their potential relevance to human brain evolution…is the single most important ‘unreported’ (or at least, unpublicized) story of the decade. I predict that mirror neurons will do for psychology what DNA did for biology: they will provide a unifying framework and help explain a host of mental abilities that have hitherto remained mysterious and inaccessible to experiments." (Ramachandran, 2001)
"We achieve our very subtle understanding of other people thanks to certain collections of special cells in the brain called mirror neurons. These are the tiny miracles that get us through the day. They are at the heart of how we navigate through our lives. They bind us with each other, mentally and emotionally." (Iacoboni, Mirroring People, p. 4)
The immediate objections were to the trumpeting of the importance of mirror neurons prior to the discovery of supporting evidence, as well as to the use of the word "miracle" to describe something that's supposed to be science. These objections ran through the seminar, much of which confronted the issue of whether or not mirror neuron claims are scientific at all.

This first objection is closely related to the first red flag of Robert Park's list of "Seven Warning Signs of Bogus Science" (2003)--that a claim is pitched directly to the media (or general public) rather than to scientists, and that specific objection was raised about Iacoboni's talk, that he was making grandiose claims to a general (or "naive") audience. This has been a common issue raised in defining the boundary between science and non-science, traceable at least back to the debate between anatomists and phrenologists in Scotland in the early 19th century, where "anatomists accused phrenologists of relying on popular opinion to validate their theories while ignoring opinions of scientific 'experts'" (to quote sociologist of science Thomas Gieryn's 1983 paper on "Boundary-work and the demarcation of science from non-science," p. 789). While it wasn't stated in this case that mirror neuron advocates are appealing to the general public to the exclusion of scientists, they were explicitly criticized for their appeals to the public in order to raise interest in their work, make it easier to get funding, and so forth, and, in the case of Iacoboni's book, for using different language in his book aimed at a popular audience that eliminated qualifiers and wasn't appropriately skeptical.

In my opinion, Iacoboni shouldn't be faulted for popularization of his work or his generation of excitement and funding from public interest--the former criticism seems a bit like sour grapes--but only for the latter, any cases where he presents arguments without proper supporting evidence, or fails to identify theoretical speculation as such. What should be significant is not the mere fact of public appeal, but the extent of the gap between the scientific evidence and the public description. Note that there will always be a gap between evidence and any scientific theory, even where a theory is firmly established, since scientific theories are always subject to further revision--they're not logical proofs. "Tiny miracles," though--I have to agree that's over the top.

Another objection raised to mirror neurons is the wide variety of human behavior that they've been proposed to explain (from the presenters' slides):
"Since their discovery, mirror neurons have been invoked to explain imitation, speech perception, empathy, autism, morality, the appeal of porn, sports team activities, social cognition, self-awareness, yawning, mind reading, action understanding, altruism, etc."
A list of neuroimaging studies purported to provide evidence for a human mirror neuron system was shown, and the question asked was how many of these studies looked at both observation of an action and execution of an action? The answer was very few, likely because observation is much easier to test in an fMRI machine compared to execution. Of those, which found evidence of activation for both observation and execution? The answer was only a single study (Gazzola, et al., 2007).

Further questions raised for discussion (from slide):
  • Is there any conceivable way to falsify MN theories?
  • As Iacoboni claimed, MNs are not anatomically-defined, and can fire in response to the same, similar, and opposite observations/actions. The whole brain, therefore, comprises the MN system. How is that useful?
  • Many researchers have moved away from hypothesizing about “mirror neurons” to “mirror systems.” Must mirror systems necessarily be composed of mirror neurons?
  • If not, are mirror neurons the most parsimonious explanation for [insert favorite behavior here]?
  • Can you generalize findings from one species to the next when one of the species possesses cognitive capabilities that have never been demonstrated in the original species? Yes, this is a “monkeys don’t have language, nor do they imitate”-based question.
  • Can individual neuron activity logically be used as an explanation for higher-order cognitive abilities?
  • How do mirror neurons handle sarcasm?
And, though not on the slide, the following claim was noted:
Similarly, a baseball pitcher’s windup is chock full of similar kinetic clues that can activate the batter's mirror neurons and help him predict the kind of pitch he will get. "This may help explain the fact that a great pitcher, Babe Ruth, was also one of the greatest home run hitters of all time," writes John Milton in Your Brain on Cubs.
and the question was asked--if mirror neurons activation is involved in imitation, rather than a complementary activity in this case, why wouldn't the mirror neuron activation interfere with Babe Ruth's ability to hit, rather than improve it? (The answer, it would seem to me, would be a suggestion that his pitching knowledge would allow him to recognize cues about the type of pitch before it happened, that would produce a benefit in hitting performance--but this is a more abstract description that doesn't necessarily require a mirror neuron explanation--another common theme of the discussion.)

This led to a lively discussion, and it seemed to me that the following were some of the most significant arguments, with my commentary on them:

1. It seems highly implausible that single cells are involved in mediating or controlling this behavior, and neither transcranial magnetic stimulation (TMS) nor fMRI is capable of isolating individual neurons. It is particularly implausible that single cells are implicated regarding a relationship to an action that is similar in that it's directed to the same goal (i.e., a semantic property). I agree, but I'm not sure why mirror neuron advocates should be taken as insisting that single cells are involved, as opposed to the "mirror neuron systems" described in Iacoboni's talk.

2. If assemblies of cells are involved instead of individual cells, how is this a distinctive or interesting theory? Doesn't it then just become a restatement of "neurons work together in the brain to make things happen?" Several people (including the person who asked that question) noted that it's still potentially interesting if these assemblies participate in both observation and action, and may provide support for theories that implicate motor programs of speech generation in speech perception.

3. Some other evidence for mirror neurons from TMS experiments in speech production. Two parts of the speech motor cortex, one active when people produce labial phonemes, and another part active when they produce dentals, were stimulated with TMS in the form of a double-pulse, which tends to provide a stimulative effect similar to priming. The result was that double-pulsing the region associated with labials facilitated the perception of labials, and double-pulsing the region associated with dentals facilitated the perception of dentals.

4. The inference to mirror neurons from fMRI evidence is choosing a single possible explanation without sufficient discriminatory evidence to exclude other explanations, such as priming. This seems like a quite reasonable objection, but one which doesn't preclude further research both within a mirror neuron framework and from outside--a battle between camps is probably a good way to provoke fruitful experimentation and mutual criticism until discriminatory evidence or arguments are obtained. There was some disagreement in the discussion about whether such discriminatory evidence could ever be obtained, but I'm inclined to think that someone will come along and provide some strong reasons to prefer going down one path rather than another.

5. In the cases where only a single or very few cells are measured, isn't that "a colossal sampling error"? One response was that the single-neuron measurement studies may have recorded from as many as 200 neurons, of which 75 showed mirror properties, of which 2/3 showed mirror properties in general (i.e., they matched individual actions and related actions directed at the same goal) and 1/3 only showed activation in response to the same exact action. I think this still presents a significant sampling issue in that there are likely hundreds of thousands of connections implicated for each neuron; I'm also a bit wary of the claims of mirror properties for related actions, where the relations may be semantic rather than simple associations, as there seems to be a potential for creative interpretation in determining what counts or doesn't count as related. That's independent of the implausibility of such properties at the individual neuron level.

6. The mirror neuron evidence and arguments seem to be like a cartoon version of science being presented to scientists and to the public (a criticism that explicitly excluded the original monkey studies). The use of the term "mirror neuron" seems like "a romantic notion that's taken on a life of its own," even though it is descriptive--you see someone else performing the same action, it's like looking in a mirror.

7. This is an unusual case in which, rather than psychology observing a behavior and theorizing neurological activity, the concept has been derived from observed neurophysiological behavior and "pyramided up," presenting challenges for theory comparison. Other competing theories don't have neural-level predictions. Are mirror neuron theories even falsifiable?

The seminar was closed with another quote from Iacoboni's book, from the end: "Mirroring People also ends on a hopeful note, the hope that science and scientific thinking may play an important role in our society."

I found it a fascinating discussion to observe, especially as issues came up pertaining to the norms of science and the demarcation between science and non-science, where scientists often appeal to criteria such as Karl Popper's falsifiability criterion. Most philosophers of science today agree that there is no sharp boundary between science and non-science (though there are certainly things that are clearly science or clearly not science), that the falsifiability criterion doesn't provide such a demarcation (and isn't strictly feasible given the nature of background assumptions and clusters of propositions involved in theory testing), and that the Mertonian norms of science are more of an ideal than reality. Science is a bit messier than that, and it seemed that some of the social aspects of "boundary-work" were in play in the discussion.

UPDATE: I should note that there were two papers of recommended reading for this discussion, which were:

Gregory Hickok, "Eight Problems for the Mirror Neuron Theory of Action Understanding in Monkeys and Humans," Journal of Cognitive Neuroscience 21:7, pp. 1229-1243 (2008).

Giovanni Buccino, Ferdinand Binkofski, and Lucia Riggio, "The mirror neuron system and action recognition," Brain and Language 89 (2004) 370-376.

I didn't get a chance to read those before the seminar, but may update this post with further comments after I do.

UPDATE (September 5, 2013): Alison Gopnik piece on "Cells That Read Minds? What the myth of mirror neurons gets wrong about the brain" on Slate.


Jane Drake Brody said...

Thanks for your clear discussion of the critique of MNs. As a theatre director and teacher, the concept of such brain activity is exciting and explains much of what we have been observing over the years. I have no doubt that such things exist, perhaps this research will take another turn, but scientists are discovering what actors have know for years.

Brad said...

As Jane mentions, everyone is aware of imitative behaviors. Mirror neurons seems like a way to market biological explanations for our human experiences.

Based on your review, it does seem that Iacoboni's main idea is to popularize his research. I don't think he really expects to find a single neuron with new distinct organelles that contrast with other typical neurons.

If I 'member right, a neurons main function is to send and receive a small number of ions, so unless those engrams are firing around invisibly I'm not sure what they hope to discover by proposing the term mirror neuron or mirror neuron system. It reminds me a bit of the hyped up evolutionary psychology inferences.

Since modeling or empathizing with someone involves input from all senses, it would follow that imitation forms from various qualia from multiple senses.

Then what would be the mirror neuron? The senses would form some sort of short term memory from sensory input, and activate behaviors in an attempt to represent outwardly what was perceived. Then would the mirror neuron be the decision to represent the behavior versus not represent the behavior?

Ian' may find that he is seeking to translate learning and attention theories into neurology.

mattyyo said...

My two cents on Mirror Neurons:

Gallese et al.(1996) found neurons in area F5 in the monkey cortex (a premotor area coding for hand and mouth) firing when both the action was observed, and when the action was made.

17% of neurons examined in the F5 area were termed ‘mirror neurons’(MN).

MNs required interaction between object and agent (hand or mouth), not just object or agent.
Matching observation of meaningful action and execution.

Responses to objects or tools to imitate action was weak or non-existent.

Mirror neurons proposed as internal representation-understanding consequences of action, so i guess maybe as Brad said "some sort of short term memory from sensory input"

Umilta et al. (2001).
Conditions were:
Full Vision - reach hand to pick up block.
Blocked vision – show the monkey the block then pull screen to block view of object then reach with hand to object.
Mimed Full Vision nothing there and reach towards nothing with hand.
Mimed Blocked Vision- show the monkey nothing there then pull screen and reach.

Half of the "MNs" found fired to both full vision and blocked vision but not the others. authors propose this infers knowledge and consequences of the action (action recognition).
"The mirror system transforms visual information into knowledge"

Baldissera et al. found that in humans, the spinal cord contained an inhibitory mechanism preventing the execution of an observed action- so one can observe an action and react to it without externally making the movement (so when someone does an action, your premotor cortex also does the same neural firing but you dont act it out).
This was thought to be an analogue of the monkey studies.

Jackson et al. (2005)

in humans: activation in the ACC and insula during the perception and assessment of someone else’s pain (areas of STS- superior temporal sulcus, also has connections with amygdala).These areas are the same for when you yourself experience pain.

Dapretto et al. (2006).

high-functioning children with autism and matched controls underwent fMRI while imitating and observing emotional expressions.

both groups performed the tasks equally well, children with autism showed no "mirror neuron activity" in the inferior frontal gyrus (pars opercularis- near insula)compared to controls. Activity in this area was inversely related to symptom severity in the social domain.

lack of MNS activity in ASD children may indicate that part of the problem of social deficits is due to dysfunctions in this system

so its gone from monkeys viewing stuff and creating an "internal representation" to "wow in humans fMRI shows activation of the SAME areas both viewing and performing a task....this must show mirror neuron circuits".

an interesting jump but one that is kinda plausible..... obviously nothing to really put all your money on but an area to investigate with further research....
again does this show "MN systems"??? probably....all it means though is that we use the same neural networks for both perceiving and doing stuff.....this makes sense b/c of the need for the brain to be an efficient processor.....