buylotrogold

The following is part two of a series of articles originally posted in September, 1991.  

If learning can be understood independently of the internal structure of the organism and quite general laws of learning can be specified, then it must be that all actual organisms implement the same learning algorithm (at least at some appropriate level of abstraction). But, if that's true, it's miraculous, given the enormously different selective pressures that different species are subjected to. For example, look at all the different ways in which different species have solved the problem of obtaining energy. Shouldn't we expect a similar multitude of learning algorithms? Where have I gone wrong here? Also, I'm surprised that there is so much energy spent here defending behaviorist learning theory. Are there actually any real behaviorists left? All the modern textbooks on learning theory I've ever read (Dickinson, Contemp. Animal Learning Theory; Mackintosh, Cond. & Assoc. Learning) are quite explicit that learning theory has evolved into a COGNITIVE theory of how animals learn about correlations. Note, these are books written by hardcore animal learning researchers and not by softies. If anyone knows of a real behaviorist currently studying animal learning, could they please let me know who they are and maybe give me a list of some of their recent publications. Thanks. Jon

Are there actually any real behaviorists left? Sure.  See JEAB, JABA.

Why is it that pigeons in a Skinner box are asked to peck a key or push a treadle instead of pressing a bar; why are the rats not asked to peck a key? Because it doesn't make any difference.  You could have either do either, it's not out of their physical competence.  The particular operant chosen is not critical.  As has been demonstrated thousands of times. You really have to be careful when making such huge generalizations like this. You can get an animal to do almost anything in an operant chamber, but how well the operant response you *force* the animal to perform can be used to research operant conditioning is questionable. Often this entails punishing the animal for an incorrect response (e.g.  Lashley jumping-stand).  You can get a rat to bar-press in one or two training sessions (about 1 hr), but you cannot get a pigeon to do it in that amount of time

The particular operant chosen is not critical. As has been demonstrated thousands of times. You really have to be careful when making such huge generalizations like this. Well, I should qualify not critical to what