Dear Yehouda Harpaz,
     
     I am afraid that this is no use to me. If you want to send a 
     presubmission enquiry, that's fine, but I am not going to waste time 
     looking up information on web sites and such like. If you want us to 
     consider this properly you are, as I indicated yesterday, going to 
     have to have to send a fully referenced abstract of the paper with a 
     letter explaining (somewhat less breathlessly than below) exactly what 
     it is you think it is you have discovered. I should say, however, that 
     we use exactly the same kinds of reviewers as Current Biology, and on 
     the basis of the limited information you have given me already I think 
     it unlikely that we would interested in considering your paper for 
     publication.
     
     Sincerely,
     Dr Rory Howlett, Nature
     
     
     


______________________________ Reply Separator _________________________________
Subject: Submission to nature about Howard et al.
Author:  Yehouda Harpaz <yeh_harlequin.co.uk> at Internet
Date:    21/01/97 13:08


     
     
     
Dear Rory Howlett, 
     
I talked with you on the phone on monday, about submission to nature. The full 
text of what I want to submit attached below, but you need a background.
     
Non-invasive brain imaging (PET and fMRI) are supposed to give us new insights 
into the brain. However, while in some cases it gives useful results, in many 
cases the results are not useful: they are irreproducible, and cannot be 
interpreted within current theories. In these cases, there is a string 
tendency to overinterpret the results. Moreover, the researchers in the field 
by now accept these overinterpretations, and suppress discussions of their 
validity. 
     
The original article by Howard et al ian example of overinterpretation. My 
comment highlights some of the important assumptions that they made, and explain
why making these assumption is not reasonable. Added to the fact that the data 
is irreproducible, as the authors themselves admit, and the conclusions that 
they present become vary unlikely. I submitted this to Current Biology. 
     
The deputy editor sent it to a review. The 'reviewer' recommended not to 
publish, based on a review of my text. However, reading his 'review' made 
it is clear that he did not read most of my text. The deputy editor did not 
accept
this, but I contacted the editor and he agreed. he sent it to another 
reviewer. The second reviewer, which was aware of the events up to this point, 
wrote a much longer 'review', in which he made false statements, misquote my 
text, and other demagogical tricks, and recommend not to publish. Based on 
this review, the editor decided not to publish. When I tried to argue he said 
he is not an expert, and must accept the opinion of the reviewer. 
     
I contacted some of the editorial board of Current Biology, and the only 
answer that I got was that I must accept the editor judgement, and suggested 
that I try another journal. 
     
Both the reviews and my comments about them are available at 
http://human-brain.org/zeki-index.html
     
The problem is that the 'reviews' that I got are representative of the opinion 
of the researchers in the field, which reject any discussion of these 
assumption, which makes their results much less striking. Thus if my text goes 
to a reviewer from the field, it is likely to be rejected using nonsense 
arguments like the reviews that I got in Current Biology. Thus the editorial 
person that deal with it must be aware of this, and be prepared to check if the 
review he gets back is actually a sensible one. Without this, it
is a waste of time.
     
     
I appreciate that this make it hard work, but the questions I raise are 
important, and must be discussed. openly 
     
     
Yehouda Harpaz
     
Harlequin Ltd.
Barrington Hall,
Barrington
Cambridge CB2 5RG
UK
     
yeh_harlequin.co.uk
     
http://human-brain.org/
     
--------------------------------------------------------------------
     
Functional Specialization in the visual Cortex?
     
     
     
     
Howard et al [1] showed their subjects three pairs of different kind of 
stimuli, and use fMRI to measure the activity in their brain. Based on their 
results, they conclude that there is "extraordinary degree of functional 
specialization in the visual cortex". However, the conclusions do not follow 
from their data. There are three major flaws in the logic of the argument that 
Howard et al [1] use:
     
     
     
 1) The authors ignore the higher level nature of the stimuli, i.e. their
conceptual and emotional significance.
     
     
     
 In higher level, the stimuli vary on many dimensions, e.g how 
 realistic they look, how interesting they are, how attractive they 
 are. Thus any difference in activity may be attributed to 
 differences in the significance of the stimuli, rather than to the 
 way they are processed.  These effects must be controlled in some 
 way, before it can be concluded that the activity seen is wholly a 
 result of low-level processes.
     
     
     
2) The authors assume that high differential activity is equivalent to 
functional significance.
     
     
     
     
     
 Howard et al [1] measured differential metabolic activity, which 
 presumably also means differential neuronal activity. However, 
 differential neuronal activity does not entail functional 
 significance. In other words, the processing of the input is not 
 necessarily done where the highest differential activity is seen.
     
     
     
 The equivalence of differential activity and functional 
 significance is widely assumed, presumably because it simplifies 
 the analysis. However, there is no evidence for it at the 
 resolution in which Howard et al[1] present their data, and it is 
 logically implausible. This is because the bulk of the information 
 in the brain is not coded in level of activity (how many neurons 
 are active in a specific region), but in the pattern of activity 
 (which neurons are active and when). Thus processing input (or any 
 other process) means changes in pattern of activity, not level of 
 activity. Hence, regions where the processing take place does not 
 necessarily appear as peaks of differential activity. On the other 
 hand, peaks of differential activity can arise from other reasons, 
 including (a) high-level processes, (b) direct effect of the input 
 (rather than processing it), (c) completely irrelevant effects. 
 Thus the equivalence cannot be assumed, and must be supported by 
 evidence.
     
     
     
     
 As the authors themselves point out, the area of differential 
 activity in their data seems to correlate with the size of the 
 visual input. On the other, it does not seem to correlate to the 
 amount of processing that is required for each stimulus. Thus the 
 data itself seems to hint that the differential activity in this 
 case is a direct result of the visual input, rather than anything 
 to do with processing it.
     
     
     
     
3) The authors regard localization as specialization.
     
     
 `Specialization`, when applied to body organs, normally means some 
 physiological modification. Even if it will be shown that specific 
 kind of stimuli are processed in a specific region, it does not show 
 that this region is modified in a special way for processing this 
 kind of stimuli. Thus localization is not equivalent to 
 specialization.
     
     
     
 It can be claimed that localization is a weak form of 
 specialization, but this is a word game that would just lead to 
 confusion. Moreover, it is clearly not what Howard et al [1] mean, 
 as they claim in their conclusions that "there is an extraordinary 
 degree of functional specialization in the visual cortex".
     
     
     
     
The patterns of differential activity seen by Howard et al [1] do not match any 
patterns seen in 
any other investigation, using any method, including, as the authors themselves 
say, their own 
previous studies. This argues against specialization. As the argument for 
specialization is 
invalid for any of the three flaws listed above, their data does not support 
specialization.
     
     
     
In summary: The data presented by Howard et al[1] shows some degree of 
localization of 
differential activity when subjects are presented with different stimuli. It 
does not show that this 
localization is not a result of higher level effects, it does not show that this
activity corresponds 
to functional significance, and it does not show any evidence for 
specialization, as opposed to 
localization.
     
     
     
References:
     
     
     
1.  Howard RJ, Brammer I, Wright I, Woodruff PW, Bullmore ET, Zeki S: A 
direct 
demonstration of functional specialization within motion-related visual and 
stodgier 
cortex of the human brain. Current Biology  1996, 6:1015-1019.