This repository contains the demo codes for replicating results in our paper: Majima K, Sukhanov P, Horikawa T, Kamitani Y (2017) "Position information encoded by population activity in hierarchical visual areas". eNeuro (in press).
The fMRI data for five subjects are avairable from http://brainliner.jp/data/brainliner/Position_Decoding. Those should be included in the "data" folder. The fMRI data were saved according to the BrainDecoderToolbox2 format https://github.com/KamitaniLab/BrainDecoderToolbox2.
In our study, a ball randomly moving in a two dimensional field was presented to human subjects as a stimulus (see the above figure in this page). The stimulus position was predicted (decoded) from brain activity by two decoding algorithms, maximum likelihood estimation (MLE) and support vector regression (SVR). While the first method provides straightforward interpretation given accurate voxel response models, the latter method is expected to perform model-free information retrieval from brain activity data.
Please run
- "exeRFMestimation_XXXXXXXX.m" in the "RFMestimation" folder
- "exeDecodingWithRFM_XXXXXXXX.m" in the "decodingWithRFM" folder
in this order. The results (true and decoded stimulus positions) would be stored in the "results" folder in the decodingWithRFM folder. Then, please run
- "makeFigDecodingAccuracyWithRFM_XXXXXXXX.m" in the decodingWithRFM folder.
The results figure would be created.
Please run
- "exeDecodingWithSVR_XXXXXXXX.m" in the "decodingWithSVR" folder
- "makeFigDecodingAccuracyWithSVR_XXXXXXXX.m" in the same folder
in this order. The decoding results would be stored in the "results" folder, and the results figure would be created.
Copyright (c) 2017 ATR Department of Neuroinformatics
The scripts provided here are released under the MIT license (http://opensource.org/licenses/mit-license.php).
Please contact us (kamitanilab.contact@gmail.com) if you have any questions.