Send to

Choose Destination
Sci Rep. 2016 Jun 2;6:27416. doi: 10.1038/srep27416.

Single-trial prediction of reaction time variability from MEG brain activity.

Ohata R1,2,3, Ogawa K1,4, Imamizu H1,3,5.

Author information

Cognitive Mechanisms Laboratories, Advanced Telecommunications Research Institute International (ATR), Keihanna Science City, Kyoto 619-0288, Japan.
Graduate School of Frontier Biosciences, Osaka University, Suita, Osaka 565-0871, Japan.
Department of Psychology, Graduate School of Humanities and Sociology, The University of Tokyo, Hongo 7-3-1, Bunkyou-ku, 113-0033 Japan.
Department of Psychology, Graduate School of Letters, Hokkaido University, Kita 10, Nishi 7, Kita-ku, Sapporo, 060-0810 Japan.
Center for Information and Neural Networks (CiNet), National Institute of Information and Communications Technology and Osaka University, Suita, Osaka 565-0871, Japan.


Neural activity prior to movement onset contains essential information for predictive assistance for humans using brain-machine-interfaces (BMIs). Even though previous studies successfully predicted different goals for upcoming movements, it is unclear whether non-invasive recording signals contain the information to predict trial-by-trial behavioral variability under the same movement. In this paper, we examined the predictability of subsequent short or long reaction times (RTs) from magnetoencephalography (MEG) signals in a delayed-reach task. The difference in RTs was classified significantly above chance from 550 ms before the go-signal onset using the cortical currents in the premotor cortex. Significantly above-chance classification was performed in the lateral prefrontal and the right inferior parietal cortices at the late stage of the delay period. Thus, inter-trial variability in RTs is predictable information. Our study provides a proof-of-concept of the future development of non-invasive BMIs to prevent delayed movements.

[Indexed for MEDLINE]
Free PMC Article

Supplemental Content

Full text links

Icon for Nature Publishing Group Icon for PubMed Central
Loading ...
Support Center