Thank you Vaughan at Mind Hacks for saving me another long-winded rant about this subject. Vaughan takes an article in the NYTimes to task for suggesting that sarcasm "resides" in particular areas of the brain:
Finally, to say that sarcasm "resides" in one particular place in the brain from a study like this is just daft. It's like concluding fun "resides" in a particular area of the city because you've noticed that the population enjoyed themselves less after the cinema burnt down.
Read the whole thing.
Mental processes don't live anywhere. They are derived from the systematic interaction of many brain regions each performing a specific information processing task. Likewise, abstractions like sarcasm -- even if they could be boiled down to discrete mental processes...which I assert they can't -- don't live in the parahippocampal gyrus or anywhere else for that matter.
While I certainly don't have a qualified background here, what about disorders such as aphasia caused by a stroke? Certainly, there are enough neurological disorders to indicate specific regions responsible for specific functions. Why would mental abstractions be any different?
Here is the problem: lots of things can cause a deficit in the ability to speak (an aphasia). You could lose you hearing. You could lose the connection between the parts of the brain that process hearing and the parts that produce a coherent response. You could lose the memory for lexical grammar. All of these produce a thing that could loosely be called an aphasia. But that does not mean that hearing lives in any of the particular locations that cause these deficits.
Neuroscientists attribute processes to locations using something called dissociation experiments. In a dissociation experiment, you look at two separate processes and see if you can find two different brain regions that cause selective deficits to for these processes. Process A could be associated with region 1, Process B could be associated with region 2, etc.
What you would like to say that is that A lives in 1 and B lives in 2. The problem with this logic is that with any two processes different enough to be distinguished, we would expect these dissociations to exist for at least part of the brain. At some point in the brain, two different processes diverge. However, this does not suggest that these two processes are fundamentally different. Merely it suggests that we have dissociated them at some point in the system.
What is in play here is really two ways to understand the brain. The first assign functions to specific regions. This view is essentially an extension of phrenology. The second views the brain in terms of systems that overlap a lot. A process doesn't live anywhere. What lives in brain regions are particular information processing functions that do not have analogs in terms like sarcasm.
Also, as an aside, I object to the description of sarcasm living in the parahippocampal gyrus because such a statement suggests that is its function. The parahippocampal gyrus does many, many things. Sarcasm isn't the only one.
Or it's like saying that the motion of an automobile resides in the motor. Or is it in the wheels, or maybe in the foot of the driver on the accelerator pedal?
In neurophysiology, showing a simple dissociation is not sufficient for proving that a function is critically dependent on some structure; you have to provide a more complex demonstration, called "double dissociation". It's not sufficient to show that the function is affected when you manipulate the structure, you have to show that it's not affected when you manipulate any other structure.
At best, Rankin's study demonstrated only a single dissociation, which doesn't really prove anything.
Thank you for taking the time for such a detailed explanation.
To be fair to Rankin et al, they performed a regression analysis on segmented grey matter concentration maps and performance on their measure of sarcasm detection. Having read the abstract from the conference (as others have noted, there's no article yet), nevertheless a regression approach is a more powerful use of the range of data than the comment by Dean Loomis would suggest.
To the degree that the size of a structure matters, demonstrating that grey matter concentration in a structure is linearly correlated with performance on a task isn't exactly "not proving anything".
However, after reading the abstract that Mind Hacks posted, I see some additional problems with the study. For one, only one patient group was actually impaired on the task (representing 10 out of the 90 or so subjects they scanned). The NYT article gives the impression that all patients were impaired. A better way to describe the article is that a host of right hemisphere regions (they did find more than just the phg) were correlated with individual differences (and not failure) in understanding sarcasm. Fair enough!