Neuroanatomical information gets interesting once we have independent evidence on the function of a brain region, and when we also know something about its engagement in a network of regions. Colin Phillips group's research on the neuroanatomy of sentence comprehension, led by Ellen Lau, Assistant Professor, University of Maryland, started when they found that the same sentence-level linguistic manipulation elicits activity in quite different brain regions in fMRI and MEG studies. And they independently knew something about region identified by the MEG studies. Getting to the bottom of that puzzle proved far more interesting than they could have imagined. And the detective work relied crucially on Ellen’s cognitive expertise. (Lau, Phillips, & Poeppel, 2008, Nature Reviews Neuroscience.)

We’re fortunate to have access to multiple brain recording tools (EEG, MEG, MRI). But we use those tools sparingly in our research. The best way to understand the brain is not always to observe brain activity — that is only informative if we have a good understanding of what we’re looking at. Effective cognitive neuroscience must be tightly linked to cognitive and computational research, and that is why we do all of them in parallel. Many of the most interesting neuroscience questions about encoding and accessing linguistic information are better investigated using computational modeling approaches for the foreseeable future, and it is not certain that current brain recording tools will be able to contribute to those discussions. I have been using cognitive neuroscience measures since the early 1990s and I have invested much effort in building infrastructure in this area, but I get frustrated by the increasing rush to focus on neuroscience research at the expense of less sexy but cost effective linguistic and cognitive research.

Colin Phillips
Department of Linguistics