Wednesday, September 24, 2014

Mirror neurons do not resonate with actions: two pieces of single-unit evidence

Gergley Csibra (2007) noticed a fundamental flaw in the logic of the mirror neuron theory of action understanding:
[there is] a tension between two conflicting claims about action mirroring implied by the direct-matching hypothesis: the claim that action mirroring reflects low-level resonance mechanisms, and the claim that it reflects high-level action understanding. The tension arises from the fact that the more it seems that mirroring is nothing else but faithful duplication of observed actions, the less evidence it provides for action understanding; and the more mirroring represents high-level interpretation of the observed actions, the less evidence it provides that this interpretation is generated by low-level motor duplication.  
Csibra's tension is evident in two famous and influential monkey mirror neuron studies by the Parma group.  In one, Umilta et al. 2001, mirror neurons were shown to fire during the observation of object-directed actions even when the object was hidden from view during the actual reaching (it was shown to the monkey before being hidden).  These same cells tended not to fire when there was no object behind the occluding screen (an empty platform was shown to the monkey).  The result was interpreted as evidence that mirror neurons are responding to the meaning or goal of the action, not the action itself. They write,

In order to activate the neurons in the hidden condition, two requirements need to be met: the monkey must “know” that there is an object behind the occluder and must see the hand of the experimenter disappearing behind the occluder. It appears therefore that the mirror neurons responsive in the hidden condition are able to generate a motor representation of an observed action, not only when the monkey sees that action, but also when it knows its outcome without seeing its most crucial part (i.e., hand-object interaction).
But if the monkey already knows the outcome, what is the point of simulating it with a motor response?  Note that the action, the movement itself, is the same in both cases.  What causes the mirror neurons to fire, then, is not the movement, but the presence or absence of the object (in the monkey's memory).  It is not a simple resonance.

The other experiment is even more famous, the report on parietal lobe mirror neurons in Science by Fogassi, et al. 2005. In this experiment, monkeys observed an experimenter reaching for an object and either bringing it to the mouth or placing it in a container. The figure here, from a 2006 Scientific American piece by Rizzolatti, shows a schematic of the set up:



The figure shows the response of a grasping-to-eat mirror neuron responding both when the monkey performs the action (1) and when he observes the action (2).  Notice, though, the the point in time when the cell starts firing during observation: it is during the reach toward the object, prior to the grasp (the red line=moment of grasp).  This is puzzling because the reach toward the object is identical in both conditions.  Only the subsequent movement, bringing to mouth or placing in container, distinguishes the action. How could the mirror neuron know ahead of time?  Is it mystically reading the mind of the experimenter?  Not at all.  

The figure is actually misleading.  If you look in the online supplemental material of the original report in Science you find this:

The container was present only in the trials in which the grasped object was subsequently placed into it. The presence of the container acted as a cue allowing the monkey to predict the most likely subsequent motor act.
It was the visual context that told the monkey ahead of time what was going to happen.  So again like the occlusion study, what determines mirror neuron activity is not the action itself but the monkey's broader understanding of the action context.  

Monkeys do not understanding because their motor system resonates with observed actions. The motor system resonates with observed actions because the monkey understands already what's going to happen.  

References

Csibra, G. (2007). Action mirroring and action understanding: An alternative account. In P. Haggard, Y. Rosetti & M. Kawato (Eds.), Sensorimotor foundations of higher cognition. Attention and Performance XII (pp. 453-459). Oxford: Oxford University Press. 


Fogassi, L., Ferrari, P. F., Gesierich, B., Rozzi, S., Chersi, F., & Rizzolatti, G. (2005). Parietal lobe: from action organization to intention understanding. Science, 308(5722), 662-667.

Hickok, G. (2014). The Myth of Mirror Neurons. New York: Norton


Umiltà, M., Kohler, E., Gallese, V., Fogassi, L., Fadiga, L., Keysers, C., & Rizzolatti, G. (2001). I know what you are doing. a neurophysiological study. Neuron, 31, 155-165. 

4 comments:

William Matchin said...

Just a side note - there is some cross-talk on the book over at Faculty of Language - http://facultyoflanguage.blogspot.com/2014/09/mirror-mirror-in-brain.html - perhaps it would be interesting to bring some of this discussion there.

ikbol said...

Greg,

I'm trying to understand you - I get the impression that you're being generally very literal - please correct me - that you treat mirror neurons as literal mirrors - "rigiform" mirrors. And also you're very focussed on "imitation."

The term that embodied cog. sci. uses is *simulation*.

I'd suggest it helps more to think of the brain's conceptual system as a simulation system - an infinitely *flexiform* simulation system.

So when you see s.o. taking action, you know that sometimes the form of the action is to be taken literally, and as far as possible *rigiform* mirrored - if say, you are trying to literally copy Charlie Chaplin's walk, or if you want to understand how Charlie Chaplin feels emotionally.

But, most of the time, you know that what you see can be simulated with great latitude. For example, if you (or a dog) see a man pointing straight to a ball a few 100m away, you (& dog) know that you do not move literally straight to the ball - you may well have to follow an extremely zig-zag path depending on the lay of the land. The straight line of his arm can be simulated very flexiformly.

Ditto if you see s.o. moving to shake your hand, you know that your hand does not have to conform precisely to his, but you have great freedom (within constraints) how to grasp.

If we think of them rather as "simulation neurons" - then precise rigiform mirroring of actions is an extreme, if fairly common, instance of what is mainly flexiform simulation of the world.

Even apparently rigiform simulation of other bodies will actually be flexiform, anyway - because their bodies are not exactly like our own, and we always have to make some flexiform internal adjustments even if we're trying to mirror them precisely.

I get the impression then that talking about mirror neurons is a major distraction. Really we're talking about the brain's conceptual system - when you/the monkey see another monkey, you're conceptualising them as well as forming a quasi-photographic image of them.

The real issue is how does the brain's conceptual system/ simulation system work?

With all the flexiformity of plasticine (hence your ability to recognize an infinite diversity of "plasticine shapes").

Pace Barsalou: "[the] concept [is] a dynamic system capable of producing an infinite number of simulations that each represent relevant information about the
category in a specific situation (e.g., Barsalou, Breazeal & Smith, 2007 )"

Somehow I would think no present neuroscientific investigations are able to pick up all the complexity of the conceptual system's operations - or indeed be sure of exactly what is going on in any brain region.







Greg Hickok said...

So the whole brain is a simulation? That doesn't solve any problems then. We are back to square one. It reduces to the how-does-the-brain-work problem.

ikbol said...

Huh? How do you get to that position? You seem to have a fondness for extreme caricature.

Here's how it probably works. When your brain sees or thinks about a hand grasping, it forms a rough, flexible figure (an outline) of that hand . [You can call it figure/image schema/map/outline]. That figure is then linked to your hand both for the purposes of simulation and understanding. Your brain/system configures your hand accordingly - and depending on context can do so from a literal/rigiform to a very flexiform way.

It is the flexiformity of your hand - its capacity to form infinite configurations - that gives your brain the power to flexibly form (and, say, draw) infinite figures/shapes of hand.

If you want to get an insight into this process look at:

http://en.wikipedia.org/wiki/Dance_(Matisse)#mediaviewer/File:La_danse_(I)_by_Matisse.jpg

From those v. simple, v. schematic figures/outlines, you can get up and dance like those dancers - do a reasonable imitation. That's very remarkable because those are actually *still* pictures - and gives you some idea of the complexity of visual/simulational.configurative reasoning that must be involved.

Looking at that Matisse must involve "mirror" neurons - but whatever those neurons (and associated networks of neurons) are doing is going to be helluva lot more complex than rigiform mirroring.

Figures/outlines are central to the brain configuring the body for action. (If you think there's an alternative to the brain using figures to process and imitate Matisse's dancing figures, pray tell what you think it is).

From my brief reading of some of the neuroscientific literature on mirror neurons, I get the impression that v. few people are taking into account the role of the conceptual system - and its figures/schemas - in mediating between seen image and action.

There wouldn't be so much simplistic analysis in terms of MN's being genetic or associative, and comparisons with behaviourism.

P.S. I hope that we wouldn't disagree that the brain can use the body for simulation purposes - *without* physically moving the relevant parts. Hence, when injured, you can know that if you move your foot in a certain way, it will hurt - even though you haven't yet moved the foot or felt any pain.