Thursday, April 7, 2011

The new semantic hub: the posterior middle temporal gyrus

Most of us agree that conceptual information is represented in a broadly distributed network throughout cortex, but there is disagreement about what the organizational principles of this knowledge might be (see debates between Alfonso Caramazza and Alex Martin or Friedemann Pulvermuller), as well as a debate about the system, or "hub", that binds all of this information together. Here I'm going to focus on the latter question.

One hypothesis is that the anterior temporal lobe serves as the brain's semantic hub (Patterson, et al. 2007). The evidence for this claim comes primarily from semantic dementia, a degenerative condition in which patients have debilitating semantic deficits that seems to cut across language, visual objects, and hearing; i.e., it looks like an amodal conceptual semantic deficit. The neural degeneration is particularly evident in the anterior temporal lobes (among other regions). If you were at the Neurobiology of Language Conference last year in San Diego, you heard a very lively debate over this issue by Carolyn Patterson and Alex Martin, which raised several questions for the ATL=semantic hub position.

An alternative position has been suggested previously on the basis of stroke data, namely, that the posterior middle temporal region is critical for some form of lexical semantic integration. For example, we argued that this region serves as a sound-to-meaning interface (Hickok & Poeppel, 2004) or in other terminology, as a "lexical interface" (Hickok & Poeppel, 2007). Why? Because damage to that part of the brain is associated with (primarily) semantic comprehension deficits (Dronkers, et al. 2004; Bates, et al. 2003).

A new paper by Turken and Dronkers (2011) adds important new information to this debate and goes so far as to elevate the MTG to the status of a "semantic hub". They used publicly available DTI and resting state fMRI datasets to map the fiber tracts and functional connectivity of several ROIs define on the basis of Dronkers et al. (2004) lesion study. The basic result was that the MTG ROI was found to be wired up, tract wise, to a broad network including STS/AG, STG, and frontal area BA 47 (the functional connectivity map shows similar areas, see below, bottom portion). This same set of regions have been independently identified by Jeff Binder's group as being involved in semantic processes on the basis of a recent meta-analysis of functional imaging studies (see below, top portion) (Binder et al. 2009).

What's interesting to me about these images is that the "semantic" areas seem to form a ring around the lower-level auditory-motor areas (STG <-> PT <-> pIFG), suggesting a hierarchical organization associated with phonological (lower-level) and semantic processes (higher level). It is worth mentioning as well that the connectivity pattern of the MTG was bilateral, both in the fiber tracing and functional connectivity analyses.

The ATL ROI was more sparsely connected both in terms of fiber tracts and functional connectivity.

An analysis of the connectivity of the STS region showed that it was largely connected to "dorsal stream" structures, but also, importantly, connected with the MTG.

In general, this strikes me as a fairly nice confirmation of the broad organization laid out in the dual stream model (HP, 2004, 2007), but I might be biased. I'm interested in your thoughts!


Bates, E., Wilson, S. M., Saygin, A. P., Dick, F., Sereno, M. I., Knight, R. T., et al. (2003). Voxel-based lesion-symptom mapping. Nat Neurosci, 6(5), 448-450.

Binder, J. R., Desai, R. H., Graves, W. W., & Conant, L. L. (2009). Where is the semantic system? A critical review and meta-analysis of 120 functional neuroimaging studies. Cereb Cortex, 19(12), 2767-2796.

Dronkers, N. F., Wilkins, D. P., Van Valin, R. D., Jr., Redfern, B. B., & Jaeger, J. J. (2004). Lesion analysis of the brain areas involved in language comprehension. Cognition, 92(1-2), 145-177.

Hickok, G., & Poeppel, D. (2004). Dorsal and ventral streams: a framework for understanding aspects of the functional anatomy of language Cognition, 92 (1-2), 67-99 DOI: 10.1016/j.cognition.2003.10.011

Hickok, G., & Poeppel, D. (2007). The cortical organization of speech processing Nature Reviews Neuroscience, 8 (5), 393-402 DOI: 10.1038/nrn2113

Patterson, K., Nestor, P., & Rogers, T. (2007). Where do you know what you know? The representation of semantic knowledge in the human brain Nature Reviews Neuroscience, 8 (12), 976-987 DOI: 10.1038/nrn2277

Turken AU, & Dronkers NF (2011). The neural architecture of the language comprehension network: converging evidence from lesion and connectivity analyses. Frontiers in systems neuroscience, 5 PMID: 21347218


Greig de Zubicaray said...

We found both left ATL and posterior MTG/STG in our paper published last year and no BA 47, though our focus was on amodal conceptual processing. Note that our study presented voxelwise data, so was not ROI driven per Turken & Dronkers. Interestingly, our DTI data showed the left IFOF and uncinate fibre tracts connecting these areas, also confirmed by Turken & Dronkers. The independent confirmation is nice as some reviewers of our paper had an issue with these findings.


de Zubicaray, G., Rose, S.E., & McMahon, K. (2010). The structure and connectivity of semantic memory in the healthy older adult brain. NeuroImage, 54,1488-1494. DOI:

Greg Hickok said...

There's just too many papers to keep up with! Thanks for the heads up on your pub, Greig. It looks impressive on quick glance. That ATL focus is awfully tiny though, and really really "A" in the TL.