- #1
Pythagorean
Gold Member
- 4,401
- 313
1) Original Article:
http://www.sciencemag.org/content/298/5594/824.short
2) Comment on Article:
http://www.sciencemag.org/content/305/5687/1107.3.full
3) Rebuttal:
http://www.sciencemag.org/content/305/5687/1107.4.full
Summary of above:
1) Original authors (Milo, et al) create a random distribution of neural network motifs and compare it to the distribution of networks in C. elegans. Any distribution that is significantly over represented compared to the random set is declared significant (result of evolutionary pressure).
2) Artzy-Randrup, et al say 'well, you didn't control for a true null-hypothesis, you didn't factor out spatial distribution. neurons form networks with other neurons that are spatially close.'
3) The rebuttal from Milo is that the spatial distribution of the neurons is a significant aspect of the results of the selection that results in this overrepresentation.
Just curious what the PF Biology expert opinions are. I tend to side with Milo here, but I'm naive. To me, knowing spatial distribution is a priori knowledge, just like any other. If we have a mechanism for WHY a nertwork motif is over-represented (i.e. spatial distribution) that that just gives us more insight, it doesn't negate the study.
What do you think? Is there a spandrelly story to spatial distribution of neurons that I'm not considering?
http://www.sciencemag.org/content/298/5594/824.short
2) Comment on Article:
http://www.sciencemag.org/content/305/5687/1107.3.full
3) Rebuttal:
http://www.sciencemag.org/content/305/5687/1107.4.full
Summary of above:
1) Original authors (Milo, et al) create a random distribution of neural network motifs and compare it to the distribution of networks in C. elegans. Any distribution that is significantly over represented compared to the random set is declared significant (result of evolutionary pressure).
2) Artzy-Randrup, et al say 'well, you didn't control for a true null-hypothesis, you didn't factor out spatial distribution. neurons form networks with other neurons that are spatially close.'
3) The rebuttal from Milo is that the spatial distribution of the neurons is a significant aspect of the results of the selection that results in this overrepresentation.
Just curious what the PF Biology expert opinions are. I tend to side with Milo here, but I'm naive. To me, knowing spatial distribution is a priori knowledge, just like any other. If we have a mechanism for WHY a nertwork motif is over-represented (i.e. spatial distribution) that that just gives us more insight, it doesn't negate the study.
What do you think? Is there a spandrelly story to spatial distribution of neurons that I'm not considering?