SciPost Submission Page
Normative approaches to neural coding and behavior
by Ann Hermundstad
This Submission thread is now published as
Submission summary
| Authors (as registered SciPost users): | Ann Hermundstad |
| Submission information | |
|---|---|
| Preprint Link: | scipost_202401_00027v1 (pdf) |
| Date accepted: | April 18, 2024 |
| Date submitted: | Jan. 22, 2024, 11:59 p.m. |
| Submitted by: | Hermundstad, Ann |
| Submitted to: | SciPost Physics Lecture Notes |
| for consideration in Collection: |
| Ontological classification | |
|---|---|
| Academic field: | Physics |
| Specialties: |
|
| Approaches: | Theoretical, Computational |
Abstract
These are a brief set of lectures notes for lectures given at the Les Houches summer course in Theoretical Biological Physics in July 2023. In these notes, I provide an introduction to some of the theoretical frameworks that are used to understand how the brain makes sense of incoming signals from the environment to ultimately guide effective behavior. I then discuss how we can apply these frameworks to understand the structure and function of real brains.
Published as SciPost Phys. Lect. Notes 83 (2024)
Reports on this Submission
Report #3 by Anonymous (Referee 3) on 2024-4-9 (Invited Report)
- Cite as: Anonymous, Report on arXiv:scipost_202401_00027v1, delivered 2024-04-09, doi: 10.21468/SciPost.Report.8855
Strengths
- Pedagogical presentation of normative approaches in neural coding.
- Clear presentation of the fly's navigation system.
- Beautiful figures and illustrations.
Weaknesses
Report
Recommendation
Publish (surpasses expectations and criteria for this Journal; among top 10%)
Report #2 by Anonymous (Referee 2) on 2024-4-6 (Invited Report)
- Cite as: Anonymous, Report on arXiv:scipost_202401_00027v1, delivered 2024-04-06, doi: 10.21468/SciPost.Report.8735
Strengths
Weaknesses
When discussing RL in Sec 2.3, it'd probably be useful to refer to the chapter by Vergassola, where RL will be discussed in detail (and a similar reference from Vergassola would be needed, too).
Not crucial, but in 3.1, when discussing local excitation-global inhibition, it might be worthwhile, for pedagogical purposes, to connect this to Turing patterns and to the LEGI model in chemotaxis. Maybe more relevant, some reference to balanced networks might also be useful.
In discussion round Fig 17, how crucial is the form of ReLU nonlinearity? For a different nonlinearity, is it always possible to select J_E, such that the distinct orientations form a Goldstone mode, rather than a collection of small bumps and valleys?
Report
Report #1 by Anonymous (Referee 1) on 2024-2-29 (Invited Report)
- Cite as: Anonymous, Report on arXiv:scipost_202401_00027v1, delivered 2024-02-29, doi: 10.21468/SciPost.Report.8639
Strengths
(2) Many useful examples
(3) Well structured
(4) Highly pedagogical
Weaknesses
Report
Requested changes
I could only find some small typos:
- Line 44: "difference timescales" --> "different"
- Line 103: "they systems" --> "these"
- Line 212: "as an one" --> "as one"
- Line 310, Eq. (7): $s$ instead of $s_t$ in the denominator
- Line 341, Eq. (10): the sum should run over $\theta_t$
- Line 348: why is there an arrow over the standard deviation?
- Line 400: "in turn impacts" --> "impact"
- Line 404: "R,L"--> "H,L"
- Line 508: class "of" behaviors
- Line 796: son "of" Werner
Minor comments: - Line 280: It is clear from the context, but maybe in addition to point (4) one could say that the underlying assumption is that the stimulus distribution is known. - Line 301, Eq. (6): the reader may wonder why we take the MSE instead of the maximum a posteriori given that we know there are only two options for $\theta$.