search

KamitaniLab / GenericObjectDecoding

Demo code for Horikawa and Kamitani (2017) Generic decoding of seen and imagined objects using hierarchical visual features. Nat Commun https://www.nature.com/articles/ncomms15037.
150 stars 46 forks source link

Stimulus_id and image_index in fMRI data #17

Closed huawei-xu closed 4 years ago

huawei-xu commented 4 years ago

When I used pandas to extract training data from subject 1, I find two problem:

  1. Maybe there is a little mistake about category index. As you can see, when I sorted table by image_index, there are only 4 images in the first category and 4 images in the 151th category. image
  2. when I use tsv file to check the relationship between the training image index and the stimulus id, I find that there may be 4 images not in right order. the No.958 image in training data is the No.959 image in the tsv file and the No.1102 image in training data is the No.1103 image in the tsv file.
ShuntaroAoki commented 4 years ago

Thank you for the reporting.

We've fixed the category_index and image_index. in the fMRI data files available at figshare (version 5; https://figshare.com/articles/dataset/Generic_Object_Decoding/7387130).

We did't notice this error because we actually don't use the indexes in the analysis scripts. Sorry for any inconvenience from this error.

And so sorry for the late response.