Baseline-subtract all the cells' firing rate (baseline is 300ms after fixation and before target onset).
Split each cell's trials into the training set and the testing set (half and half, random)
Construct pseudo-trials from training set and testing set for later discrimination analysis. (An example pseudo-trial from the training set: for all N neurons, randomly pick one trial from each cell's training set, so this pseudo-trial contains data from all N neurons as if they were recorded simultaneously.) For each target location, we constructed 1000 pseudo-trials. In total, we had 1000x7 training trials and 1000x7 testing trials.
For each training time point and each testing time point, we performed PCA to reduce the dimensionality of the training data to explain 95% of the original variance (denoise the data), and projected the testing data onto the same PCs.
For each training time point and each testing time point, an LDA classifier (linear discrimination analysis, MATLAB function classify) was trained on the denoised training data and tested on the denoised testing data and the performance was reported.
Copyright: Content may be subjected to copyright.
How to cite:
Readers should cite both the Bio-protocol preprint and the original research article where this protocol was used:
Tang, C., Herikstad, R., Parthasarathy, A., Libedinsky, C. and Yen, S.(2020). Minimally dependent activity subspaces for working memory and motor preparation in the lateral prefrontal cortex. eLife. DOI: 10.7554/eLife.58154
Do you have any questions about this protocol?
Post your question to gather feedback from the community. We will also invite the authors of this
article to respond.
0/150
Tips for asking effective questions
+ Description
Write a detailed description. Include all information that will help others answer your question including experimental processes, conditions, and relevant images.
Spinning
Post a Question
0 Q&A
Spinning
This protocol preprint was submitted via the "Request
a Protocol" track.