GitHub - WilburDoz/Prefrontal_Structured_Working_Memory_Representations: Code to accompany Dorrell et al. 2026, A Normative Theory of Prefrontal Structured Working Memory Representations

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
El_Gaby_Saved_Theory		El_Gaby_Saved_Theory
Subspace_Alg_Saved_Theory		Subspace_Alg_Saved_Theory
Contextual_Compositional_Example.ipynb		Contextual_Compositional_Example.ipynb
El_Gaby.ipynb		El_Gaby.ipynb
Loss_is_not_convex.ipynb		Loss_is_not_convex.ipynb
Mushiake.ipynb		Mushiake.ipynb
P&B_Data_Analysis.ipynb		P&B_Data_Analysis.ipynb
P&B_Theory.ipynb		P&B_Theory.ipynb
Size_Schematic.ipynb		Size_Schematic.ipynb
Stroud.ipynb		Stroud.ipynb
Subspace_Algorithms.ipynb		Subspace_Algorithms.ipynb
Synthetic_Data_Regression_Testing.ipynb		Synthetic_Data_Regression_Testing.ipynb
Xie_Data_Analysis.ipynb		Xie_Data_Analysis.ipynb
Xie_Theory.ipynb		Xie_Theory.ipynb
readme.txt		readme.txt

Repository files navigation

Guide to what files do what:

Synthetic_Data_Regressison_Testing
Looks at different ways of estimating subspaces from correlated data.

Xie_Theory, P&B_Theory, El_Gaby, Mushiake, Stroud (X)
Generates theoretical representations for each of the named experiments

Contextual vs. Compositional is the beginning example that shows the axis of single neuron responses

Subspace algorithms generates the two different gating vs. conveying choices

Xie_Data_Analysis, P&B_Data_Analysis
Performs data analysis for the listed experiment

Loss_is_not_convex shows that the recurrent loss is not convex.

Size_schematic generates the simple size plot in figure 5A.