Tuesday, March 27, 2012

New Rule #1

Use of the following terms are now banned when used without qualification:

  • phonological processing
  • morphological processing
  • syntactic processing

Such terms need to be followed by phrases such as "___ in the context of task ..." or something similar. To use these terms alone is meaningless.  David tried to point this out back in 1996:


Poeppel, D., A critical review of PET studies of language. Brain and Language, 1996. 55: p. 317-351


(PET studies?  What's that?  Dude, you're old, ha!)  The point was that the pattern of neural activity varies depending on what task is used to measure "x processing".  This point was again underlined in each of the Hickok and Poeppel papers.  Yet, I still see people saying things like "Broca's area activity is strongly correlated with phonological processing ability".  That's kind of like the language equivalent of saying V5/MT activity is strongly correlated with visual processing.  Ok, true in some cases, but not in others and even when it is true the statement isn't very helpful in the context of understanding neural circuits for information processing.  

Sixteen years after David's 1996 paper (dude, you're old!) we know a little bit more about how language is organized in the brain.  The circuits that are involved in x processing vary depending on whether the task is production or comprehension based, whether the materials are written or spoken, whether the task requires conscious attention to x or is more automatic, and so on.  There's no excuse for using vague terminology anymore.  It's just muddling the field.


9 comments:

Dominik Lukeš said...

Thanks. You made my day. I think that the reputation of a journal should be measured by how much of this voodoo they publish. Sadly so many disciplines are based on the assumption that X processing a real thing. And keep banging on about how collocation of processing A with processing B (measured in completely different contexts) means that condition Y is related to cognitive process C.

How much can we trust anything in this area? A lot of it sounds vaguely plausible but then so did bloodletting and witchcraft!

Dominik Lukeš said...

Funny, just clicked on a link and found an example of how this sort of stuff can have real life consequences (Only "language skills" substituted for "language processing"):
http://blog.dinolingo.com/2012/03/04/why-girls-are-better-than-boys-in-language-learning/

Unknown said...

Sure, I agree. However, how can you design an experiment that yields results that are NOT task specific but only reveals something about the underlying mechanism (whatever "mechanism" means...)?

By the way, does the criticism of PET studies not apply to ERP studies?

- Julius F.

Greg Hickok said...

You can't and that's precisely the point. If you are interested in studying phonological processing as it is employed in auditory comprehension then use an auditory comprehension task. If you are interested in studying the neural basis of performing same different judgments on pairs meaningless syllables then use that task. Just don't do one and say the findings apply to the other.

Unknown said...

Agreed. A slightly related issue pertains to ecological validity of some some fMRI tasks (phonological and others...). I suspect that comparing activation in 'stupid task 1' to 'absurd task 2' to 'nonsense task 3' reveals nothing particularly helpful. BTW, David and YOU are old :)

- JF

Greg Hickok said...

Ya, but not as old as David! HA!

Anonymous said...

Hi Greg,

Your rule makes a lot of sense. And, of course, it's your blog!

Ok, let's take a 5 years old and simple example:
'However, the ultimate goal of these studies is presumably to understand the neural processes supporting the ability to process speech sounds under ecologically valid conditions, that is, situations in which successful speech sound processing ultimately leads to contact with the mental lexicon and auditory comprehension'.

Does the term 'ecologically valid conditions' also refer to a word/pseudoword task in which 'successful speech sound processing ultimatley leads to contact with the mental lexicon'? Also, I am not sure to really and precisely catch your definition of 'auditory comprehension' in this context? etc, etc...

I am joking of course. But I don't believe that your position ('there's no excuse for using vague terminology anymore') is always tenable.

Also, regarding the famous paper by David Poeppel, it would be fair to cite the reply by Démonet, Fiez, Paulesu, Petersen and Zatorre (Brain & Language, 1996, 55, 352–379).

Best,

Marc Sato

Greg Hickok said...

Hi Marc,
I assume you are referring to some study of mine that used a word/pseudoword task and you are pointing out that I am guilty of the same sin? If so, I would agree with you. We've all done it. The point is that we all need to stop and think about what ability or process we are trying to understand and then ask whether the task we are employing is actually tapping into that process. The empirical record demonstrates that there are many and dissociable "phonological processes" for example, so we can't simply assume that one task that we might use experimentally will tap into the processes in the task we are ultimately interested in. So if this is the case, why not just try to use the task you are interested in understanding, at least when this a feasible.

In what kind of cases would be have to be vague?

Yes, thanks for posting the response to David's piece.

David Poeppel said...

Thanks, Marc, for pointing to the reply to my old paper. My personal favorite of the comments - because of the cynical tone and nastiness - is the one by Eraldo Paulesu (with whom, incidentally, I get along very well).

For bibliographic completeness, I think I am permitted to point out that I also had a response-response, a meta-response to the comments:

Some Remaining Questions about Studying Phonological Processing with PET: Response to Demonet, Fiez, Paulesu, Petersen, and Zatorre (1996).

"One cannot but be grateful for such a detailed and spirited response to a paper (Poeppel, 1996), even when the outcome is such that one is deemed ‘‘naive,’’ ‘‘superficial,’’ ‘‘unrealistic,’’ and so on. In spite of my personal shortcomings, the article has at least served to unify several different groups using PET and motivated them to clarify the rationale underlying some of their studies."