Tuesday, July 10, 2007

The auditory dorsal stream may not be auditory

No this is not a territorial concession to that other sensory modality (although I have heard "seers," shall we call them, claim regions of the posterior planum for their beloved albeit imperialist visual system). Instead it is closer to a territorial concession to the motor system, a concession, we propose, that the see-scientists also need to make in their dorsal stream.

Some background: When we first started writing about a dorsal processing stream for speech back in 2000, we looked to the dorsal visual stream for inspiration. (Actually, the dorsal stream idea was included in our 2000 TICS paper as a response to a reviewer's comment; it wasn't in the original manuscript. But that is another blog entry.) The dorsal visual stream started out as a "where" system, but was then re-invented, on the basis of good evidence, as a "how" stream (i.e., a visual-motor integration system). In the late 1990s, a ventral/dorsal division of labor had been proposed for the auditory system as well, in which dorsal=where. Given that there were obvious needs for auditory-motor integration for speech (see any Hickok & Poeppel paper), we proposed that the dorsal auditory stream supported such auditory-motor functions analogous to the dorsal visual stream. Subsequent fMRI studies in the Talking Brains West lab identified area Spt as the auditory version of "visual" areas such as LIP, AIP, and so on.

But there's a problem with thinking about any of these areas as part of a specific sensory stream. For example, the "visual" areas are organized around motor effector systems (LIP=eyes/saccades, AIP=hands/grasping), and they can take input from multiple sensory systems. So these regions appear to be linked more tightly to specific motor modalities than to a given sensory modality: LIP is not going to get excited about grasping behaviors, but if auditory input helps guide behaviorally relevant eye movements, LIP is happy to oblige as Richard Andersen showed a few years back. Maybe, then, it makes more sense to talk about sensory-motor integration areas in the posterior parietal lobe rather than visual-motor integration areas. And maybe area Spt, our previously hypothesized auditory-motor area, is not so much auditory as it is a sensory-motor integration area for the vocal tract effector. If this is true, we can make two predictions. (1) Spt should be multisensory as long as the sensory input is relevant to vocal tract actions (somato feedback is a likely candidate), and (2) Spt should be less excited about auditory tasks when the output behavior doesn't involve the vocal tract.

My (now former) grad student, Judy Pa, tested prediction number 2 in an fMRI experiment reported in a forthcoming paper in Neuropsychologia. She had skilled pianists listen to novel melodies and then reproduce them either by covert humming (vocal tract) or covert playing (manual articulators). The critical finding was that Spt showed an attenuated response during the motor phase of the task for the playing condition compared to the humming condition. A region in the intraparietal sulcus showed the reverse pattern.

So, as David and I hinted in our Nature Reviews Neuroscience paper, current evidence suggests that the posterior parietal lobe and posterior planum contain a network for sensory-motor integration areas. These areas are not part of one sensory system, but are tightly linked to specific motor effector systems. Spt is part of this network, and is specifically tied to the vocal tract. Conclusion: Spt and the posterior portion of the planum that it occupies is not part of the auditory system. But then neither is LIP or AIP part of the visual system. Does this mean that it doesn't make sense to talk about dorsal streams at all then? Not sure.


Anonymous said...

There is a new paper "Sensory Systems: Auditory Action Streams?" by Timothy D. Griffiths. He wrote about some new psychophysical studies support evidence that actions streams in auditory domain could be connected with control of voice fundamental frequency. What do You think about this?

Greg Hickok said...

Sounds like an interesting paper. The idea sounds reasonable, and is not in conflict with my claim that the auditory dorsal stream may not be exclusively auditory.