In a comment to my last post on the CNS Summer Institute, TB Down Under rep, Greig de Zubicaray, mentioned Indefrey and Levelt's 2004 paper as an example of a solid attempt to merge neuroscience and psycholinguistics. I'm my response comment, I agreed, but cautioned that we don't actually learn much by localizing neural correlates of the various boxes in an assumed psycholinguistic model, as Indefrey & Levelt did. Such localization exercises don't tell us how language is processed in the brain, only where it is processed, and "where" by itself, isn't that interesting. Seriously, who really cares if phonological encoding in speech production involves the dorsal or ventral bank of the STS (for example)? But, as Greig points out, "where" has the potential to reveal "how," and I agree.
But what have we learned? We are more than decade removed from the first PET and fMRI studies of language processing. Presumably this is enough time to assess progress, so this seems like a good time for an exercise:
What are the Top 10 contributions of functional imaging (PET & fMRI, specifically) to understanding how language is processed in the brain? [crickets]
We better get some feedback on this question, because if the neuroscience of language community can't come up with anything on this one, it probably means we are wasting our time generating pretty pictures. Go ahead and draw on your own work.
Greig suggested that some of his own imaging studies indicate that the architecture for language processing system are more closely related to connectionist architectures than serial feedforward architectures.
Can we come up with a Top 10 list?
Let me ask this: Can you come up with a single fMRI study contributing a key insight into language representation?
Well, I'm a pretty big fan of this 2006 Journal of Neuroscience paper with Brian Gold and David Balota on semantic priming. Since they had a real psycholinguist on the project, the materials and design are great and they did all the controls you're supposed to do. They show a clean anatomical distinction for the famous short-SOA/automatic vs. long-SOA/strategic contrast--MTG in both, but frontal only in long SOA, as you might expect--which they replicate across three experiments. The directionality of those effects, which they test with further contrasts, is even more interesting and begins to give us some insights about how different parts of LIFG might contribute differentially to strategic priming effects, consistent with other work on LIFG from Anthony Wagner's group.
As a psycholinguist, I'd love to see more fMRI papers like this that take a phenomena that has an extremely well-understood suite of behavioral correlates, and bring it to fMRI without screwing with crucial aspects of the materials or timing of presentation. Not to mention having an Experiment 2 where you replicate and extend your finding!
Nobody is responding to this for a reason: There aren't any.
One of the sources of the problem we are having in the field is that the neuroscientists don't know much about linguistic processes and the psycholinguists know very little about the brain. And both categories understand little about complex systems dynamics. On the one hand, we get neuroimaging papers from researchers trained as psycholinguists. These papers are very weak on the brain imaging methodology and interpretation. In many case, they are almost useless. On the other hand, we also get neuroimaging papers by neuroscientists. These papers tend to be better on the neuroimaging methodology by very simplistic on the theory. Why aren't these researchers collaborating? On the surface they are, in some cases. In practice, big egos despise collaborations. This is especially true of older generation, I feel. The current academic system still favors the lone genius protective mentality and model. Until true collaborations take place, I don't see how any neuroimaging papers of language will have any impact or reveal anything of substance.
This lack of response makes me optimistic for the prospects of my own career, or/and pessimistic about the people following this blog.
Since everyone seems to be in love with the mirror neuron, maybe the few ostensibly following this blog with approval are masochistic, in love of the venerable mirror neuron, and unwilling to accept the less painful truth.
The fourth stage of grief over the pervasive misunderstanding of how to achieve valid inference from fMRI and Pet? The age of functionalneurorelativism?
I think that the softball answer is that there are plenty of reasonably well replicated findings of functional localization suggesting the dilation, constriction or otherwise information of the conceptual space in which all sorts of linguistic theories should be responsibly formulated. I'll leave the how and why for a wider audience. In the mean time, lets all please learn how to conduct decent studies( following the moderators' examples).
I hope that the answer to "will we one day completely disregard the vast majority of fMRI and Pet data collected before 2010, and achieve a level of collective understanding that makes it impossible to publish the crap that is currently representative of most of our field?" will be yes!
Unfortunately, the trend is accelerating. High impact factor journals more and more publish articles only based on who's the author. Are they big cheese or not? Will their name result in the popular media give us publicity by mentioning this article? Foundations function in a similar manner. Will the hot shot we give to, provide publicity for our foundation/cause? National grant agencies work in a similar manner.
We have all seen some hot shots give a talk, say at the CNS meeting over the last two years, and think: "Why the heck is this person considered a hot shot? Why was this person invited to give a symposium talk? The theory is silly, the data is confounded, and it's not even an entertaining or well-organized talk!" Obviously, these individuals were shamelessly promoted by their mentors, regardless of scientific talent. Mostly, with the purpose of creating a little army of loyal soldiers who are going to protect their back on grant agencies panels and on editorial boards.
Post a Comment