Wednesday, February 9, 2011

Why is Broca's aphasia/area the focus of research on "syntactic comprehension"? Was it a historical accident?

Arguably it was the classic paper by Caramazza and Zurif, published in 1976, that kicked off what turned into decades of research on the role of Broca's area in syntactic computation. We all know from our grade school lessons that Caramazza and Zurif found that Broca's aphasics exhibit not only agrammatic production, but also a profound deficit in using syntactic knowledge in sentence comprehension. The critical bit of evidence was that Broca's aphasics seemed perfectly fine in using semantic information (lexical and real-world knowledge) to derive the meaning of an utterance: they could correctly understand a so-called semantically non-reversible sentence like, The apple that the boy is eating is red. But they failed when interpretation required the use of syntactic information, i.e., when the sentence was semantically reversible like, The boy that the girl is chasing is tall.

This finding suggested that Broca's aphasics had a deficit in syntax, one that affected both production AND comprehension. Broca's area, via its association with Broca's aphasia (a dubious association, but a topic for another post) then became the major anatomical focus for the localization of syntax, including its role in comprehension of sentences. This obsession with Broca's area and syntax (comprehension in particular) persists today.

But is it all a historical accident? I happened to re-read Caramazza and Zurif today (I'm working on a chapter on Broca's aphasia and it is always a good idea to go back to original sources). C&Z tested not only Broca's aphasics, but also conduction aphasics, a little-remembered fact. Conduction aphasics have posterior lesions and don't have agrammatic speech output. But guess what? C&Z report that the conduction aphasics performed exactly like the Broca's aphasics. Check out the graph below which I recreated by eye-balling the relevant values from their Figure 3 which shows percent correct on the sentence-to-picture matching task for object-gap semantically reversible vs. nonreversible sentences like the examples above.

The failure of conduction aphasics to use syntactic knowledge was, of course, noted by the authors.

...the conclusion is inescapable- Broca’s and Conduction aphasics do not seem at all capable of using algorithmic [syntactic] processes. Thus, for those sentences that were semantically constrained, performance was approximately at the 90% level, but it dropped to chance level when these semantic constraints were not available. p. 580

Why then did all subsequent work focus on Broca's aphasics and Broca's area? Why was conduction aphasia and more posterior lesion sites not considered as a possible test case/source for the neural substrate for syntax?

The answer derives from the common interpretation of conduction aphasia at the time, which is that of a disconnection syndrome. Conduction aphasia was caused, the story went, not by damage to any computational system, but by a disconnection of computational systems, namely Wernicke's and Broca's area. C&Z argued that the conduction aphasics comprehension problems derived from a disconnection of syntactic systems, which lived in Broca's area.

Conduction aphasics also were incapable of using syntactic algorithmic processes [see also Saffran & Matin (in press) and Scholes (in press)]. The question arises, therefore, as to whether syntactic operations also rely on cortical regions posterior to Broca’s area or whether the conduction deficit should be considered within a disconnection framework, that is, as the severing of a connection to Broca’s area (Geschwind, 1970). Given the impressive arguments offered by Geshwind, we are presently satisfied in treating it as a problem of disconnection, but a disconnection from an area that subserves sytactic processes. p. 581

But the interpretation of conduction aphasia has evolved since in the 1970s. It is no longer considered a disconnection syndrome but rather a deficit caused by cortical dysfunction. We can, and should, argue about what conduction aphasia is, functionally. Maybe our final conclusion will be the same as C&Z's (I don't believe it will), but the point is that based on an assumption about the nature of conduction aphasia, research emphasis shifted entirely to Broca's aphasia and Broca's area, ignoring conduction aphasia and more posterior cortices. I believe this was an unfortunate and ultimately misleading turn.

Maybe historical accident isn't the right word for what happened in the 1976 publication. It wasn't an accident that C&Z assumed the popular account, articulated so eloquently by Geschwind. It was a reasonable conclusion. But it did dramatically shape the focus of subsequent research and we are still living with the consequences of this theoretical argument. There are still heated debates both in print and in conference forums regarding the role of Broca's area in syntactic comprehension (Grodzinsky & Santi, 2008; Rogalsky & Hickok, 2011; Willems & Hagoort, 2009). Conversely, there is no concerted effort aimed at determining the role of the temporal-parietal junction (the location of lesions associated with conduction aphasia) in syntactic comprehension. This is a shame because I believe we are missing a big part of the puzzle.

This is a good lesson though. Sometimes ideas become entrenched in the scientific literature by "accidents" of the current theoretical milieu, and sometimes the resulting scientific path that a field takes can be the wrong one. It is important to occasionally re-visit the reasons why choices were made and to evaluate whether a new direction is worth exploring.


Caramazza A, & Zurif EB (1976). Dissociation of algorithmic and heuristic processes in language comprehension: evidence from aphasia. Brain and language, 3 (4), 572-82 PMID: 974731

Grodzinsky Y, & Santi A (2008). The battle for Broca's region. Trends in cognitive sciences, 12 (12), 474-80 PMID: 18930695

Rogalsky C, & Hickok G (2010). The Role of Broca's Area in Sentence Comprehension. Journal of cognitive neuroscience PMID: 20617890

Willems RM, & Hagoort P (2009). Broca's region: battles are not won by ignoring half of the facts. Trends in cognitive sciences, 13 (3) PMID: 19223227


Tobias (Freiburg) said...

Hi Greg,

my impression from my clinical work with aphasic subjects is that most of them, irrespective of their lesions, have problems with semantically reversible passive sentences. I think there exist a few studies on this but I would be tempted to run a comprehension study on subjects with only discrete symptoms. Even the fluent subjects with posterior lesions (very mildly aphasic, former "Wernicke's aphasics") seem to have difficulties with these sentences. I would expect even control subjects do.

So I consider it a highly unspecific symptom not limited to lesions in the left IFG.



Greg Hickok said...

Hi Tobias,
I agree completely which makes it all the more puzzling that Broca's area remains the dominant focus, right?