Friday, July 10, 2009

Somatotopic organization of action word meanings - Again!

Some interesting commentary on motor somatotopy (or lack thereof) from Greig de Zubicaray's website which I've copied and pasted here (with permission of course):

We're very interested in claims of a so-called "semantic somatotopy" in primary and premotor cortex (i.e., overlapping activity for executing/observing actions performed with a specific effector and retrieving word meanings related to that effector). See Tash Postle's paper in NeuroImage last year in which the peak maxima from neuroimaging studies cited as evidence for this claim were plotted against cytoarchitectonic maps of BA4 and BA6 in the language dominant left hemisphere. Friedemann Pulvermüller and colleagues have a new fMRI paper using cluster analysis in press in Human Brain Mapping in which they conclude that "The motor-premotor region was dominated by semantic somatotopy... with face, arm, and leg word activation foci at inferior lateral (-49/11/16), dorsolateral (-32/-5/52), and dorsal sites (-19/-21/61)." So, here are 10 mm spheres centred on those peak maxima plotted in relation to the cytoarchitectonic maps of BA4 (red) and BA6 (blue) from Eickhoff et al. (2006). Again, not much congruency with respect to the actual cortical motor areas, and as motor execution and/or observation conditions weren't included in the study, we have no evidence that these clusters are representative of the actual neuronal ensembles used by the participants to perform effector movements. Interestingly, though, the leg peak is in BA6 this time.


Pulvermüller, F., Kherif, F., Hauk, O., Mohr, B., & Nimmo-Smith, I. (in press). Distributed cell assemblies for general lexical and category-specific semantic processing as revealed by fMRI cluster analysis, Human Brain Mapping.

Eickhoff, S.B., Heim, S., Zilles, K., & Amunts, K., (2006). Testing anatomically specified hypotheses in functional imaging using cytoarchitectonic maps. NeuroImage, 32, 570–582.


Joe Devlin said...

I'm not sure what your point here is. Are you suggesting (via Greig) that somatotopic activations related to action words are a false claim or just that this study isn't particularly good?

Seems to me that it is very difficult to make a strong somatotopy claim without collecting data in the same individuals showing actual somatotopy in motor representations and then showing a correspondence between those regions and the action words that activate them (or not, as the case may be). But haven't previous groups done just that?


Greg Hickok said...

Greig's point is that the foci of activation in the "somatopic activation" are not in motor cortex. This is a problem if you want to claim, as Pulvermuller is inclined to do, that the meaning of action words are coded in motor cortex. I'm sure Greig can elaborate...

Greig de Zubicaray said...

I think Greg has summarised the issue nicely. Tash Postle evaluated the neuroimaging evidence cited in support of the 'semantic somatotopy' hypothesis in relation to cytoarchitectonic maps of motor areas, finding little congruency with actual motor areas (BA4 or BA6) or between studies as part of her PhD thesis. In order to support this hypothesis, some minimal criteria need to be satisified: activity associated with effector action words should be localised in motor or premotor cortex and arranged in a somatotopic manner corresponding to activity elicited by actual effector movement. Put simply, I guess we are saying "near enough is not good enough" in relation to the evidence presented.


Postle, N., et al. (2008). Action word meaning representations in cytoarchitectonically defined primary and premotor cortices. NeuroImage, 43, 634-644.

Joe Devlin said...

Right, so I think we'd all agree that this particular study isn't great for the reasons mentioned. But it sounds like you guys want to make a broader claim and discount the whole notion of 'action somatotopy' for words, is that right?

If so, then several other studies need to be explained away, and I'm not certain cytoarchitecture maps are the best way to do this. For instance, Hauk & Pulvermuller's (2004) paper did localise leg, hand and mouth motor actions and compared them activation for related action words. To my mind, it doesn't matter whether those activations matched the cyto maps of 10 different subjects or not -- if there was a correspondence between the somatotopy of Hauk's subjects' motor activity and action word activity, that is compelling.

Similarly, there is TMS evidence that also needs to be explained. Pulvermuller et al (2005) used TMS of primary motor cortex to localise a hand area of M1 and a leg area (based on evoked MEPs). They then show that stimulation of those two areas had differential effects on hand and leg-related action words. To my mind, this is the stronger evidence than the imaging work because it localised the M1 area in each individual and showed that it affected action words in that individual.

For what it's worth, I don't have any agenda here -- I'm just curious to understand the perceived issue. Obviously a weak study is not helpful in the debate because it essentially offers no additional information. It doesn't advance or harm the main theoretical claim - it just doesn't add anything at all.

Just my $0.02

Greg Hickok said...

Hi Joseph,

I *would* like to discount the whole notion of "action somatotopy". I think the idea is wrong. But you are right, we need to look at the whole constellation of evidence, not just whether or not activations line up with cytoarchitectonic maps.

Here's why I think an extreme version of somatotopic semantics is wrong (or more strongly, can't be right): the same motor action can be associated with multiple action meanings and multiple motor actions can be associated with the same action meaning.

Consider the action of pouring water from a pitcher into a glass. This motor action can mean POUR, FILL, EMPTY, SPILL (if the water misses its mark), DEFY (if the actor was forbidden to perform the action), etc., etc. Alternatively I can FILL by pouring water from a pitcher (using any number of particular motor actions or even my feet!), dipping a cup into a lake, turning a faucet, leaving a cup out in the rain. It doesn't take more than about 60 seconds of thought to realize that a true motor theory of action semantics (where the meanings are encoded in specific motor effector actions) is essentially bankrupt as a theory. Action concepts HAVE to be coded in a much more abstract format. Perhaps this is in frontal action-related cortex, but it can't be in M1.

So where do all the somatotopic effects come from? How about simple association? Read the word 'kiss' and what do you think of? Or put differently, what does it prime? Besides frogs, it probably primes motor actions associated with kissing which may show up in functional activation studies, etc.

Greig de Zubicaray said...

Hi Joseph,

In her paper, Tash provides a nice summary of the methodological issues in the Hauk et al. (2004) paper and others as well as the rather equivocal results from TMS studies.

Friedemann Pulvermuller said...

From Friedemann Pulvermuller, Cambridge


You miss the point of our HBM paper. Our study for the first time shows using an obective data driven method that the semantic circuits, the semantic cell assemblies if you wish, are distributed - and even more distributed than many of us dared to think before.

The Hickok Poeppel model does not account for such distributed and category specific semantic cirucits. Maybe you should change it?

BASIC: If the voxel of *maximal activation* is slightly besides your prediction - what does this mean in the face of really huge activation clouds? You must show that the could DOES NOT REACH INTO the premotor and motor cortex. Look at David Kemmerer's recent plot: Indeed some find somatotopic semantic activation outside the predicted fields - but most studies show good corresponence between motor areas and action/body-semantics.

BTW, as previous contributors have emphasied, we have used motor localiser and the finger movement area. Hand word activations do indeed overlap. The mapping to standardized motor maps is indeed irrelevant here.