reproducible research in the neurosciences

[falvarez at desire.wright.edu]

I had follow this thread with lot of interest.
I do neuroanatomy and the size of my image data sets makes this approcah 
impracticable. I see more use when numerical data is involved. Unfortunately 
numerical data from image capturing and analysis equipment is _not_ raw data.
However even in my field there is some relevance to the proposition. Papers 
in top ranked journals have to be small because very tight space limitations. 
One major constraint is in the number of figures. This has provoked a 
continuoulsy increasing use of the "data not shown" strategy for many of the 
morphological findings reported in Science, Nature, Neuron.... Unfortunately 
the "data not shown" are usually controls and other data needed to justify the 
interpretation of the data shown. In my experience the frequency of 
unreproducible or not-exactly reproducible data can be alarmingly high among 
this "data not shown" statements. It would be usefull to have a place where 
the authors have to show the nature of _all_ their data even when the journal 
did not have enough space to publish it.

Francisco J. Alvarez
Assistant Professor
Department of Anatomy
Wright Sate University
Dayton, Ohio.