What it is:

A representational similarity analysis of single-trial EEG data from a numerosity change-detection task.

What we did:

  • Used linear decoding on single-trial EEG from a numerosity oddball task with quantities 1-6.
  • Built whole-epoch and time-resolved representational dissimilarity matrices.
  • Compared PI boundary models, ANS, and control models, including a no-1 analysis.
Archived preview of EEG numerosity RSA figures

Archived preview of the MDS and RDM figure files currently available in this project.

Read Full Project Details...

This project came from our EEG work on numerical cognition and asks how different numerosities are represented relative to one another in the neural signal. Rather than focusing only on averaged responses, I wanted to work from single-trial EEG and compare the structure of the condition relationships across the full pattern of activity.

I used pairwise linear decoding and representational similarity analysis to build both whole-epoch and time-resolved neural RDMs from an oddball change-detection task with numerosities 1 through 6. The analysis compares competing PI and ANS accounts, includes control models for reaction time and pixel area, and repeats the main tests with numerosity 1 removed.

For now I am keeping this page at the project-summary level rather than posting the full analysis and interpretation here. Once our write-up is ready, I plan to link the paper or preprint so readers can see the full motivation, methods, and results in the proper publication format.