Talk:Hartline-Ratliff model
Editor's first thoughts
Dear Professor Knight,
Thank you for your fine article. I've put some suggested changes into it as reviewers' comments; I hope these help. It might also be good to break up the article into subsections, such as "History", "Static model", "Dynamic model", and "Conclusions".
Cheers, Robert.
Robert P. O'Shea, Editor, Vision; Professor and Head of Psychology, School of Health and Human Sciences, Southern Cross University, Hogbin Drive, Coffs Harbour, 2450, Australia phone: +61 2 6659 3313; fax: +61 2 6659 3202 www: <http://www.researcherid.com/rid/C-5723-2008>
Reviewer A
This is a good article. I only have a few small comments/suggestions: 1. I believe the article will be easier to read if it is divided into subsections: steady state, Limulus retina, etc. 2. Figure 1 is hard to read. It may strongly benefit readers not familiar with this system to have a more detailed caption, explaining what we are looking at at different parts of the micrograph.It would also help to overlay boxes and arrows in color - they will be more visible. Finally, I do agree with the suggestion to accompany the figure with a schematic, showing just the boxes and the connections.
Reviewer B
This article is very clear and illuminating. I have only a few suggestions for improvements.
1. In equation 1 you suppress the inhibitory thresholds--I would guess for ease of presentation. But from a physiologist's point of view the thresholds are interesting and plausible. Would it be acceptable to present eqn 1 as a linearized version of the "true" H-R equations which apply in the steady-state case, and then show that they disappear when you consider the dynamical case? Or tell the reader that you will develop the formalism for the linearized equations but that for the real eye you may need to add the thresholds?
2. Echoing the other reader, I suggest an additional figure to supplement Figure 1 to illustrate the neural circuit you have in mind. Let me suggest you use Figure 1 in Brodie et al 1978b--the block diagram of the Limulus eye.
3. You chose to emphasize the linearity of the H-R equations but in my own opinion another important property of H-R is the recurrent nature of the circuit--and this is one reason that the H-R equations, and Limulus, are so relevant to present-day models of the cerebral cortex: the recurrency. The recurrent architecture is also why you pick up the 1/1+K term and that could be clarified in the article.
4. My preference would be not to use the symbol "f" in equation11, but to use "I" already there-- to me f is confusing with the cap F of the spatiotemporal transfer function that will come soon.
5. Part of the rationale for comment 4 is that I believe that the concept of the spatiotemporal transfer function F(q, omega) is of fundamental importance (pun intended). I think the spatiotemporal transfer function deserves more emphasis as the object of analysis and the key to understanding how the network works. Also, here again, in the special case of Limulus, the spatiotemporal transfer function has the structure it does because you are analyzing a recurrent network. Cluing the reader in to why the spatiotemporal transfer function has the structure that it does, as a consequence of the network's architecture, may be pedagogically useful.
Otherwise it is perfect.
