User Tools

Site Tools


Welcome to the CoSMo 2014 hands-on session on decoding neural ensemble data!

The Tutorial modules below constitute a step-by-step walkthrough that introduces you to a data set of 100+ neurons, recorded simultaneously from hippocampal subfield CA1 as a rat runs a T-maze task, followed by some example decoding analyses.

These analyses have the potential to reveal the representational content of cognitive processes such as recall, deliberation, and planning at fast timescales.

You should feel free to play with the code parameters and start exploring at any point in this tutorial. There are some suggested exercises and questions embedded in the tutorial designed to verify your understanding of the material; you can pick and choose among these depending on what catches your interest.

If you finish the tutorial part, there are two suggested “mini-projects” that indicate directions for cutting-edge analyses you can pursue in a more open-ended way, building on the material in the tutorial but requiring more active coding and problem-solving on your part.

Finally, this is a wiki: if you see an opportunity for improvement, go ahead and edit it. These pages will stay available for at least a number of months after the course. If you'd rather not edit but just comment, there is a discussion/comment box at the bottom of each page.



  • Bieri et al. (2014) found that theta cycles with slow-gamma power (thought to reflect inputs from CA3 to CA1) were associated with sequences more forward of the animal compared to those with fast-gamma power (thought to reflect inputs from EC to CA1). This seems counterintuitive if we assume that each gamma cycle is associated with an “item” in the sequence (e.g. Lisman and Jensen, 2013): in that case, slow-gamma should have fewer, not more, items per sequence! Can we resolve this apparent discrepancy by quantifying the speed of the sequences during slow- and fast-gamma associated theta cycles?
  • This particular data set was recorded when the animal was mildly water-deprived. The maze had food on the end of one arm, and water on the other arm; a prediction of the theory that hippocampal replay reflects a goal-directed planning process is that the content of this process should be influenced by motivational state. That is, a hungry animal may “replay” food-associated trajectories more than water-associated trajectories, and vice versa. Can we test this idea by mapping the content of replays onto the maze?



Enter your comment. Wiki syntax is allowed:
analysis/cosmo2014.txt · Last modified: 2014/08/15 22:47 by mvdm