A possible functional localiser for identifying brain regions sensitive to sentence-level prosody

Abstract

Investigations of how we produce and perceive prosodic patterns are not only interesting in their own right but can inform fundamental questions in language research. We here argue that functional magnetic resonance imaging (fMRI) in general – and the functional localisation approach in particular – has the potential to help address open research questions in prosody research and at the intersection of prosody and other domains. Critically, this approach can go beyond questions like ‘where in the brain does mental process x produce activation’ and towards questions that probe the nature of the representations and computations that subserve different mental abilities. We describe one way to functionally define regions sensitive to sentence-level prosody in individual subjects. This or similar ‘localiser’ contrasts can be used in future studies to test the hypotheses about the precise contributions of prosody-sensitive brain regions to prosodic processing and cognition more broadly

Keywords:

• fMRI
• prosody
• intonation

Acknowledgements

We thank Sabin Dang, Jason Webster and Eyal Dechter for help with the experimental scripts and with running the participants, and Christina Triantafyllou, Steve Shannon and Sheeba Arnold for technical support. We thank the members of the Kanwisher, Gibson and Saxe labs for helpful discussions and the audience at the ETAPII (Experimental and Theoretical Advances in Prosody II) conference in 2011 for helpful comments. For comments on the manuscript, we thank Ted Gibson, Ben Deen, Mike Frank, Nancy Kanwisher, and two anonymous reviewers. For the script used to extract acoustic features we thank Michael Wagner. For advice on the statistical analyses of the acoustic features, we thank Peter Graff, Kyle Mahowald and Ted Gibson. We also acknowledge the Athinoula A. Martinos Imaging Center at McGovern Institute for Brain Research, MIT.

Funding

This research was supported by Eunice Kennedy Shriver National Institute Of Child Health and Human Development Award [K99HD-057522 to EF].

Notes

1. In principle, studies that ask questions like ‘does a particular manipulation activate brain region x?’ could also inform deep issues in cognitive science, but this is only possible in cases where region x is characterised sufficiently well to serve as a neural ‘marker’ of a particular mental process. With a few exceptions, most brain regions lack such detailed functional characterisation and thus are not suitable for use as markers of particular mental processes (see e.g., Poldrack, 2006).

2. Note that this experiment was not originally designed to study prosodic processing. Hence the inclusion of both meaningful (Sentences, Word lists) and meaningless (Jabberwocky, Pseudoword lists) conditions may seem unmotivated. However, we think it ends up being a strength of this experiment to be able to generalise across the presence of meaning in the stimuli: as we will show in the Results, similar patterns hold for meaningful and meaningless conditions when examined separately.

3. One important caveat to keep in mind is that individuals may differ with respect to how strongly they activate prosodic representations during silent reading. In the extreme case, if an individual activated prosodic representations during silent reading to the same degree as during auditory linguistic processing, then the contrast proposed here would fail to identify any prosody-sensitive regions in that individual. As will be shown below, the proposed contrast successfully identifies regions with the specified functional properties in the majority of individuals, suggesting that a substantial proportion of individuals have brain regions that respond more to the presence of structure in the auditory stimuli than in the visual stimuli (which we argue plausibly reflects sensitivity to sentence-level prosody). Once these prosody-sensitive brain regions are established as robust to irrelevant differences in the materials, task, etc., investigating individual differences in their response profiles and relating them to behaviour will be a fruitful avenue for future research.

4. Although in the remainder of the paper we will use the term ‘prosody-sensitive’, the reader should keep in mind that we are referring to brain regions that are sensitive to sentence-level prosodic contours.

5. The dataset used for the current study is the same dataset as that used in Experiment 3 in Fedorenko et al. (2010).

6. Not being able to define a fROI in every single subject using the fixed-threshold approach – i.e., when parcels are intersected with thresholded individual activation maps – is not uncommon. For example, when developing a localiser for high-level language regions, Fedorenko et al. (2010) considered a region meaningful if it could be defined in 80% or more of individual participants (see also Julian et al., 2012, where 60% or more of individual participants is used as a criterion for selecting meaningful high-level visual regions). An alternative that would enable one to define a fROI in every single subject would be to move away from the fixed-threshold approach. In particular, once a region has ‘established itself’ (i.e., once we know that it emerges consistently across people, is stable within individuals, and is robust to various properties of the localiser contrast), we can simply take the top – with respect to the t-values for the relevant functional contrast – 5% or 10% of voxels within some spatial constraint (defined anatomically or with the use of functional parcels obtained from a GSS analysis). This approach ensures that (1) a fROI is defined in every individual (and thus the results are generalisable to the whole population, as opposed to the proportion of the population for whom the fROIs could be defined), and (2) fROIs are of the same size across individuals (see Nieto-Castañon & Fedorenko, 2012, for a discussion). The reason that we used the fixed-threshold approach in the current paper is that it is mathematically not trivial to use the top-n-voxels approach for the conjunction of multiple contrasts, which is what we use in the current study.

Additional information

Funding

Funding: This research was supported by Eunice Kennedy Shriver National Institute Of Child Health and Human Development Award [K99HD-057522 to EF].