Friday, April 30, 2010

Auditory short-term memory and the left superior temporal gyrus

Where is the "phonological store"? Ask the typical cognitive neuroscientist on the street and you will probably be pointed to the left inferior parietal lobe. But this is incorrect. First, the idea that there is a dedicated "phonological store" is probably incorrect. Second, the system that supports the temporary maintenance of phonological information isn't in the parietal lobe, but in the superior temporal region, i.e., the same general region that supports phonological processing during speech recognition. These claims have been made on the basis of functional imaging data (e.g., Hickok et al., 2003; Buchsbaum & D'Esposito, 2008). But now there's lesion evidence to back it up.

Leff et al. (2009) studied a whopping 210 stroke patients, testing them on a range of speech tasks and their measure of auditory STM, digit span. Not surprisingly, damage to pretty much the whole left perisylvian cortex correlated with digit span measures (figure below, top row of images). But when the authors factored out processes such as speech production (naming measure) single word speech repetition (auditory word and nonword repetition) and higher-level functions (verbal fluency measure), the remaining correlate of digit span was a relatively small sector of the STG/STS (figure below, bottom row of images). This region was further shown to correlate with auditory (but not visual) sentence comprehension; it did not correlate with auditory word comprehension.

What this shows is that the left STG/STS is critical for auditory STM. What is less clear, though, is how this region relates to systems involved in phonological processing during normal speech recognition. The issue centers on whether short term maintenance of phonological information is achieved by activating the same phonological processing networks that are involved in speech recognition or whether there is a separate "store". The fact that the left STG/STS region identified by Leff et al. did not correlate with auditory word comprehension seems to suggest a separate phonological store. However, this isn't necessarily the case. For example, if phonological maintenance involves only a sub-portion of the phonological recognition network -- e.g., if the recognition system were bilateral as we've argued -- then maintenance and recognition may dissociate, so the non-correlation with auditory word comprehension is not surprising. Why does left but not right STG/STS damage cause STM deficits? Because STM is dependent on connections with the motor speech system, which is strongly left dominant.

What seems more puzzling though for a common network model is the fact that the left STG/STS region correlated with digit span even when nonword repetition was factored out. That is, nonword repetition requires accurate phonemic perception of the stimulus and interface with the motor speech system, which should implicate left phonological processing systems. So how can damage to left STG/STS affect digit span but not nonword repetition? One possibility is that the damage to left STG/STS represents partial damage to the phonological system within the left hemisphere, perhaps specifically involving phonological sub-networks that represent larger phonological chunks or sequences; or maybe it produces just enough damage to affect the more difficult processes while leaving the easier tasks spared.

Generally, this finding provides fairly compelling evidence that the left STG/STS plays a critical role in auditory/phonological STM.


Buchsbaum, B.R., and D'Esposito, M. (2008). The search for the phonological store: from loop to convolution. J Cogn Neurosci 20, 762-778.

Hickok, G., Buchsbaum, B., Humphries, C., and Muftuler, T. (2003). Auditory-motor interaction revealed by fMRI: Speech, music, and working memory in area Spt. Journal of Cognitive Neuroscience 15, 673-682.

Leff, A., Schofield, T., Crinion, J., Seghier, M., Grogan, A., Green, D., & Price, C. (2009). The left superior temporal gyrus is a shared substrate for auditory short-term memory and speech comprehension: evidence from 210 patients with stroke Brain, 132 (12), 3401-3410 DOI: 10.1093/brain/awp273

1 comment:

Anonymous said...

I like your analysis of hemispheric differences in phonological processing,to account for the lack of digit span correlation with auditory word comprehension (AWC). It seems true that the RH could be competent for AWC but not for other language functions, especially those involving articulation.
Yet I think there is another factor at work in the "missing correlation". Because all participants had to be able to do the digit span, those with severe AWC deficits were automatically excluded. Therefore this deficit could not figure in the analyses.