In re: LINGUIST List issue: http://linguistlist.org/issues/17/17-1286.html 
 
In reference to the paper claiming that Starlings may possess a uniquely 
human trait, that of recognition of recursion, I would like to point out 
that this latter conclusion is incorrect. 
 
The sequences Starlings had to respond to were of the sort AABB or AAABBB, 
for instance. This means that rather than internalizing the concept of 
recursion, all they had to do was to count. If two B's were preceded by two 
A's, then they pressed a bar and were rewarded. they didn't have to possess 
recursion for that. But even that could be argued against. The birds could 
just detect a change in pattern, say a change from A to B. If they detected 
it an adequate amount of times when it was (accidentally) 'recursive' (note 
that there were 10,000 - 50,000 trials), this would become statistically 
significant, enabling to falsely argue that they succeeded in recognizing 
recursion. 
 
The conclusion must be that Starlings do not possess anything similar to 
the core property of human language (recursion). The above comments are 
also valid to studies of Tamarins (by Hauser and Fitch). In fact, 
experiments involving self-embedding (AB)^n do not show anything about 
recursion because of this complication.