Handedness measures for the Human Connectome Project: Implications for data analysis

ABSTRACT

Open data initiatives such as the UK Biobank and Human Connectome Project provide researchers with access to neuroimaging, genetic, and other data for large samples of left-and right-handed participants, allowing for more robust investigations of handedness than ever before. Handedness inventories are universal tools for assessing participant handedness in these large-scale neuroimaging contexts. These self-report measures are typically used to screen and recruit subjects, but they are also widely used as variables in statistical analyses of fMRI and other data. Recent investigations into the validity of handedness inventories, however, suggest that self-report data from these inventories might not reflect hand preference/performance as faithfully as previously thought. Using data from the Human Connectome Project, we assessed correspondence between three handedness measures – the Edinburgh Handedness Inventory (EHI), the Rolyan 9-hole pegboard, and grip strength – in 1179 healthy subjects. We show poor association between the different handedness measures, with roughly 10% of the sample having at least one behavioural measure which indicates hand-performance bias opposite to the EHI score, and over 65% of left-handers having one or more mismatched handedness scores. We discuss implications for future work, urging researchers to critically consider direction, degree, and consistency of handedness in their data.

KEYWORDS:

• Handedness
• Edinburgh Handedness Inventory
• measurement
• laterality
• neuroimaging

Acknowledgements

This research was supported in part by grant 52935 from the John Templeton Foundation titled: “What Drives Human Cognitive Evolution?”

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data which support the findings of this study are available from the Human Connectome Project (HCP). Some restrictions apply to the availability of these data, which were used under approval of the HCP for this study. Supporting data and R code can be made available from the corresponding author, LMR, under reasonable request. However, raw handedness measures can only be provided when requests are accompanied by proof of prior permission for restricted data usage directly from the HCP (as detailed at https://www.humanconnectome.org/study/hcp-young-adult/document/ restricted-data-usage).

Notes

1 Some HCP data, including information about participant handedness, is considered “restricted” by the HCP, meaning that researchers must complete paperwork and request permission in order to gain access to that information. Information on restricted access for the HCP data can be found at https://www.humanconnectome.org/study/hcp-young-adult/document/restricted-data-usage.

2 We had to ask for the raw NIH toolbox scores for the test-retest data, which led us to an interesting discovery about the HCP’s definition of “raw” NIH toolbox data. It seems that the HCP “raw” NIH toolbox data for the 45 test-retest participants in the larger s1200 data release is actually an average of the two scores from T1 and T2, meaning that for these participants, our laterality indices (Grip strength LI and Pegboard LI) were computed on averaged “raw” data, unbeknownst to us, when the original analyses took place, whereas the remaining 1134 participant LI’s were calculated on single-measure (truly “raw”) data.