METHODOLOGICAL foundation
Toward robust and replicable cognitive neuroscience of language
Only published papers are included; for preprints, see Papers.
Last Updated: May 2024
Identifying language areas functionally in individual brains
Much early research in the cognitive neuroscience of language relied on a brain-averaging approach, where individual brains are aligned in a common space and voxel-wise correspondence is assumed. This approach suffers from low sensitivity, low functional resolution, and low interpretability. Moreover, results obtained from brain-averaged analyses are difficult to compare across studies, which is, of course, a cornerstone of a cumulative research enterprise. As a result, our group has been advocating an alternative approach where all the analyses are performed at the level of individual participants.
See Resources for FAQ on functional localization and to download localizers, parcels, an SPM toolbox for individual-subject analyses, and probabilistic functional atlases.
This paper introduces the functional localization approach in the domain of language, including developing a general algorithmic method for defining individual functional regions of interest (fROIs) by combining group-level parcels, or masks, with individual activation maps: the Group-constrained Subject-Specific (GSS) approach.
This paper extends the GSS approach to the regions in the ventral visual stream (e.g., the FFA and the PPA).
This paper introduces an auditory version of a language localizer and shows that it works similarly to the reading-based version.
This paper discusses and provides evidence for the advantages of the functional localization approach.
This paper provides a brief history of individual-subject analyses in human cognitive neuroscience and discusses their advantages.
These early papers advocate the use of individual-subject analyses in language research.
This paper discusses how the functional localization approach can be applied to research on aphasia.
Language atlas (LanA) and other probabilistic atlases
See Resources to download LanA and other probabilistic functional atlases.
Having been collecting fMRI data since ~2007 and having each participant complete a language “localizer” task, in 2022 we developed a probabilistic atlas for the language system based on individual activation maps of >800 individuals: the Language Atlas, or LanA. This atlas can serve as a universal reference frame for the language system, as it allows to estimate for any location in a common brain space (MNI or Freesurfer) the probability that it belongs to the language network, including in studies that have not included a language localizer task. This paper describes the atlas.
We have also developed similar atlases for the Multiple Demand (MD) network and the Theory of Mind (ToM) network.
Individual-level neural markers of language processing
A lot of work in the cognitive neuroscience of language attempts to relate brain activity to behavior. However, in many cases, the reliability of the neural (and, in some cases, behavioral) measures is not established. This paper shows that when the language areas are defined at the individual-subject level, some of the neural markers extracted from those areas are indeed test-retest reliable, including both across scanning runs within a session and between scanning sessions.
For an update on these analyses with larger numbers of participants, see:
Attempts to replicate influential findings from others’ work
Scientific progress is impossible if we cannot build on each other’s work. As a result, we have led or participated in efforts to replicate a number of past findings in the cognitive neuroscience of language. These include:
