Wednesday, January 26, 2011

On the relation between auditory-motor area Spt and conduction aphasia

Conduction aphasia is characterized by relatively frequent phonemic speech errors with self-correction attempts and difficulty repeating speech verbatim; comprehension is relatively well-preserved. The classical account holds that conduction aphasia is caused by damage to the arcuate fasciculus. However, we have been arguing for some time that conduction aphasia is caused by damage to area Spt -- a functionally defined region in the vicinity of the left planum temporale that exhibits auditory-motor response properties, and which we claim computes a mapping between auditory and motor speech representations, critical for aspects of speech production.

Our hypothesized link between conduction aphasia and area Spt just got stronger. In a forthcoming paper in Brain and Language Buchsbaum et al. show that the region of maximal overlap in lesion distribution of a group of 14 conduction aphasics includes area Spt (based on fMRI data from over 100 participants).

We argue that the auditory-motor transformation function carried out by Spt is necessary for verbatim repetition but also plays a critical role in internal monitoring during speech production, thus explaining the increased speech error rate when the system is damaged. This explanation does a better job of explaining the co-occurrence of phonemic paraphasias and repetition deficits than does the current dominant model of the deficit in conduction aphasia, namely, that it is a working memory deficit.


Buchsbaum BR, Baldo J, Okada K, Berman KF, Dronkers N, D'Esposito M, & Hickok G (2011). Conduction aphasia, sensory-motor integration, and phonological short-term memory - An aggregate analysis of lesion and fMRI data. Brain and language PMID: 21256582

Baldo, J.V., Klostermann, E.C., and Dronkers, N.F. (2008). It's either a cook or a baker: patients with conduction aphasia get the gist but lose the trace. Brain Lang 105, 134-140.

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.

Hickok, G., Erhard, P., Kassubek, J., Helms-Tillery, A.K., Naeve-Velguth, S., Strupp, J.P., Strick, P.L., and Ugurbil, K. (2000). A functional magnetic resonance imaging study of the role of left posterior superior temporal gyrus in speech production: implications for the explanation of conduction aphasia. Neuroscience Letters 287, 156-160.

Hickok, G., Okada, K., and Serences, J.T. (2009). Area Spt in the human planum temporale supports sensory-motor integration for speech processing. J Neurophysiol 101, 2725-2732.


Jared Medina said...

I've enjoyed reading your blog. I have an issue with a claim made in this post, namely that the lesion overlap distribution speaks to the neural substrates that underlie conduction aphasia.

A lesion overlap map identifies both regions that are associated with a deficit, but also (importantly) identifies regions that are commonly lesioned in stroke victims. Since some brain areas are more frequently damaged in stroke than others, lesion overlap maps can be misleading - confounding areas associated with a deficit and areas commonly damaged due to ischemic vulnerability.

For example, if you take the lesion overlaps of individuals with visual field deficits, you'll find that areas commonly damaged include posterior superior temporal gyrus and temporoparietal junction. But clearly, that's not the area involved in early vision. Only when you compare the regions brain damaged in individuals with visual field deficits to regions damaged without visual field deficits (using methods like VLSM) do you find that it's primary occipital cortex that's involved. (See Figure 4 in Rorden & Karnath, 2004.)

A similar point was made by Argye Hillis with regards to "apraxia of speech" (Hillis et al., 2004); previous lesion overlap studies associated speech apraxia with anterior insula damage - but didn't take into account that the insula is very commonly damaged after middle cerebral artery stroke.

Note that this is not to say that Spt is not involved in mappings between auditory and motor speech representations, conduction aphasia, etc. Just that it's necessary to be very cautious using lesion overlap data, due to the issues laid out above.

Hillis, A.E., Work, M., Barker, P.B., Jacobs, M.A., Breese, E.L., and Maurer, K. (2004). Re-examining the brain regions crucial for orchestrating speech articulation. Brain, 127 (7), 1479-1487.

Rorden, C. & Karnath, H.O. (2004). Using human brain lesions to infer function: a relic from a past era in the fMRI age? Nature Reviews Neuroscience, 5(10), 813-819.

Greg Hickok said...

Hi Jared,

Thank you for your post. I agree completely. But at the same time, I don't think your point affects the interpretation of the data we present because we effectively used the lesion data to test a specific hypothesis rather than to identify the lesion location per se.

Our hypothesis was this: damage to Spt is the cause of conduction aphasia and therefore patients with conduction aphasia should have lesions involving Spt (probabilistically defined of course). If the lesions associated with conduction aphasia did not involve Spt, then this would count as strong evidence against our hypothesis. But we found that Spt was in fact lesioned in conduction aphasia, consistent with our prediction.

I think this is a good example of how the combination of lesion and functional imaging data really help constrain interpretation of what we see in the lesion and functional imaging results alone.

Let me know what you think.

Jared Medina said...

Thanks for your response. I do agree that if very few or none of your conduction aphasics had damage to Spt, that would likely be strong evidence against your hypothesis.

However, I don't think that the lesion overlap data alone supports your hypothesis. To take the visual field cut example again, you could have an a priori hypothesis that temporo-parietal junction is crucially involved in primary vision, and the lesion overlap data would be consistent with this hypothesis. However, the lesion overlap would be misleading, for reasons I laid out in my last comment. Or, if one were to compare functional neuroimaging data on areas active for primary vision to the lesion overlap, one would either find no conjunction between the two methods, or possibly conjunction of areas on the "outskirts" of significance of both analyses - perhaps in some occipitotemporal region. Both of these would be wrong because (in this example) the lesion overlap is misleading.

However, if one were to use some sort of analysis that takes into account the lesion location of those without visual field cuts (e.g. VLSM or something similar), that analysis clearly implicates primary occipital cortex in visual field cuts, and would match up beautifully with functional imaging data on primary vision.

Bringing it back to Buchsbaum et al., the major question regarding the lesion data is this: how many subjects with Spt damage do not have conduction aphasia? If there is a large enough percentage of individuals in that category, then it casts doubt on hypotheses that Spt is necessarily involved in conduction aphasia. And this is possible as Spt is often damaged after stroke.

So, if the data taking into account subjects w/o conduction aphasia looks like this (Spt+ = damage to Spt; Aph+ = conduction aphasia):

       Spt+ Spt-
Aph+ 12    2
Aph-   3    52

Then it supports your hypothesis. However, if it looks like this:

       Spt+ Spt-
Aph+ 12    2
Aph-  38   17

...then it's a lot less clear. Hence the need for data from a population of brain damaged individuals without conduction aphasia.

Greg Hickok said...

Hi Jared,

I still agree with your general point, and agree that a +/- lesion analysis is very important generally and worth doing in our case, but I disagree with the suggestion that the data we presented does not support our hypothesis.

Let's use your example. You correctly pointed out that if had a hypothesis that TPO junction supported early vision then lesion data on visual field deficits could erroneously support this hypothesis. But, if had done our homework and looked carefully at non-lesion data on visual cortex, we should have realized that TPO junction is not a good candidate for early vision. A better hypothesis based on say functional imaging data would be a more posterior occipital distribution of early visual cortices. Now, the combination of our better-informed hypothesis with the lesion data correctly identifies the critical regions and excludes the TPO junction.

So what it boils down to how good your a priori hypothesis is. Based on a decade of research, I feel pretty confident in my hypothesis that Spt is a critical site for auditory-motor integration. If you believe that too, and believe that auditory-motor integration deficit is a good way to think about conduction aphasia, then we should agree that the lesion data support our claimed hypothesis. Of course you may want to argue that the Spt hypothesis is wrong or you may question the idea that conduction aphasia is an auditory-motor deficit... but that is a different discussion that the methodological issue your raised here.

Tobias said...

Hi Greg,

thanks for the interesting post! If I understand correctly you suggest that the same routines underlie both repetition (of nonwords, also of words?) and spontaneous speech (through monitoring). I am using a cognitive framework here.

Could this position be falsified? Would dissociations within aphasic subjects count? I wonder whether any such dissociation has been reported?

I work with an individual with deep dysphasia. She is unable to repeat nonwords and occasionally produces semantic errors in single word repetition (e.g., repeating "cabbage" as "savoy"; digit span is 1). Her spontanous speech, however, is void of any phonemic errors. Apparently, she is unable to map phonemes on motor output in nonword repetition but, on the other hand, doesn't produce phonemic errors in speaking.
Any thoughts?

Best regards,

Tobias Bormann

Greg Hickok said...

Hi Tobias,
Interesting case! The difference between repetition and spontaneous speech is that repetition requires a direct mapping between sound and motor speech systems (dependent on Spt/dorsal route) whereas sponaneous speech involves two possible routes, one mapping between a mapping between concept and motor speech and the other going from concept to auditory to motor. If the latter is broken but the former is robust you could get repitition errors without obvious errors in spontaneous speech. Where is the lesion?

Tobias Bormann said...

Hi Greg,

thanks for the reply. Our subject has bilateral lesions which included the temporal-parietal junction in the left hemisphere. However, she also has lesions located more anterior in the temporal lobe and an older right-sided parietal lesion.

You seem to suggest that there are two routes to speech output (input-output mapping via Spt and concept-output connections for spontaneous speech).

Since conduction aphasics (at least of the "reproduction variant") produce errors both in repetition and spontaneous speech, they would have lesions to both routes. Wouldn't it be more plausible to assume a functional lesion to speech output processes post lexical access (some sort of post-lexical response buffer [Morton] or phonological encoding [Levelt])? This has been suggested by Caplan and Waters (1995, B&L). A single functional lesion could explain errors in both tasks rather than two lesions which you seem to suggest.


Tobias (Dept. of Neurology, Freiburg)

Greg Hickok said...

I'm happy that this issue is being discussed. I've been thinking about conduction aphasia (off and on) for more than 10 years. It was this syndrome that really spurred me to team up with David and start working on what has become our dual stream model. There were always some features of the syndrome that didn't quite make sense, but now I think I've got it worked out. Let me know what you think.

I believe there are two routes to production, a lexical-semantic to motor route and a lexical-semantic to auditory to motor route (just like Wernicke said). I think both routes are used during normal speech production. The extent to which the dorsal sensory-motor route is needed, though, is dependent on the strength of the motor representations for a given word or phrase: if a word is common, the motor program for generating it will be coded as a chunk and will be more automatic. For less common words the motor program will have to be assembled on the fly and need more auditory guidance. Think of a typing analogy. You can type your name pretty much automatically, but try to type lateral geniculate nucleus and you may need to piece it together with a little more sensory guidance.

So damage to the dorsal, sensory-motor route can, in general, be expected to cause severe deficits on tasks that depend exclusively on this route, like non-word repetition. But one would also expect some deficits on word production, especially those words that require more sensory guidance. This is classic conduction aphasia: phonemic paraphasia is spontaneous speech + severe deficits in verbatim repetition of complex words or phrases or nonwords.

There are cases however, like yours Tobias, in which nonword repetition is impaired yet very few errors are noted in spontaneous speech. This is a bit of a puzzle. I think an important clue, though, is that these cases are quite rare, which may indicate that such patients have an atypical organization of their language system and may not be representative of the norm. Functionally, what seems to be happening is that these patients have very robust motor representations for previously learned words that can be executed accurately with little need for auditory guidance.

I about a week, I have a paper that will be appearing that lays out a more detailed model of the speech production system (the dorsal stream). I'm not allowed to talk about it yet because it's embargoed, but once it's out we can hopefully have a vigorous discussion. It affords a more detailed explanation of conduction aphasia.