The French Connection I: An important paper that Greg and I overlooked

When working on our paper for Nature Neuroscience Reviews (Hickok & Poeppel, 2007; see an early blog entry), we overlooked a terrific paper that we should have cited and the results of which we should have incorporated.

Giraud AL, Kell C, Thierfelder C, Sterzer P, Russ MO, Preibisch C, Kleinschmidt A. Contributions of sensory input, auditory search and verbal comprehension to cortical activity during speech processing. Cerebral Cortex. 2004 Mar;14(3):247-55.

Giraud and her colleagues presented participants with (i) regular sentences, (ii) broad-band speech-envelope noise signals (BBSEN), and (iii) narrow-band speech-envelope noise (NBSEN). Subjects were scanned with fMRI before a training period (only regular sentences intelligible) and after training (BBSEN intelligible, NBSEN not intelligible). BBSEN is highly intelligible after training, NBSEN remains entirely unintelligible -- see paper for details of the materials.

Anne-Lise and her colleagues were then able to separate the same physical stimulus when it could be understood (comprehension condition) versus not. This is the same intuition that forms the basis for sine-wave speech studies, some of Sophie Scott's studies, and many others (e.g. Athena Vouloumanos' experiment in J. Cog. Neuroscience).

(1) The stimulus attributes were reflected in activation in superior temporal cortex (including STS) bilaterally. (2) Natural speech compared to speech-envelope modulated noise selectively activated STS, again bilaterally. (3) Comprehension -- i.e. BBSEN after training as well as regular speech -- implicated bilateral MTG and inferior temporal areas.

So they found a convincing separation between areas principally responsible for sound analysis and areas mediating intelligible speech. Their data also enrich what the interpretation for the left STS should be. And their data show quite nicely that ventral stream areas are strikingly bilateral.

Sorry Broca-Wernicke-Lichtheim-Geschwind --- that part of lateralization is just wrong! It's much more bilateral when you look at comprehension and ventral stream contributions.

This paper is full of interesting details and discussion. If you are working on the neural basis of speech perception, language comprehension, intelligibility etc. I suggest you read this one. And -- a nice bonus -- the paper provides quite a bit of strong evidence for the model that Greg and I argued for in the 2007 paper.