Poverty of the Stimulus: Part 1 Chomsky 1959

Noam Chomsky long ago. The pen and notebook in the pocket marks the pre-PDA intellectual. (For larger photo see source.)

Download PDF of this post

Commenter Raymondw provided a reading list for behaviorist rebuttals to Noam Chomsky’s poverty of the stimulus argument. (They are here) It is a valuable list and I hope blog visitors will check it out. This week I’m going to make several posts surveying the “poverty of the stimulus” question and at the end the commenter will provide his reaction to my pontificating.

The Chomskyan revolution is commonly dated to publication of his Syntactic Structures (1957) but I date it a little later, to the 1959 publication of his review of B.F. Skinner’s book Verbal Behavior. (review here) In the 1950s Skinner was considered America’s most important psychologist. He did extensive experimental work with pigeons, training them to do many things using a method called operant conditioning, and he showed beyond the shadow of a doubt that animals could be trained to behave in many surprising ways if the behavior was rewarded. (Punishment worked less well than rewards.) What made Skinner controversial was his insistence that all learning could be explained in terms of a stimulus (feature of the environment) and response (behavior). In today’s computer world we would say input and output, but Skinner didn’t care about internal computations any more than he cared about personal thoughts. Behavior was to be explained under the Newtonian concept of cause and effect, just as the behavior of a billiard ball can be completely explained in terms of the forces striking the ball and setting it in motion.

Chomsky’s review was truly revolutionary because it overthrew Skinner’s reign as emperor. He made several arguments—such as showing that Skinner’s use of terms like stimulus and response was too vague to be scientific—but the most important was what came to be called “the poverty of the stimulus.” Chomsky, citing Karl Lashley, said (as he himself had said in Syntactic Structures):

the syntactic organization of an utterance is not something directly under the control of outside stimulation and intraverbal association, and … the syntactic organization of an utterance is not something directly represented in any simple way in the physical structure of the utterance itself.

Here we have the basic proposition of the poverty of the stimulus: you cannot determine the rules for organizing sentences simply by studying the physical organization of sentences. You cannot know from sentence structure alone that while you can say both I hit the ball and The ball hit me, you cannot say both I hit the ball he threw and *The ball hit me he threw. If you want to say that last sentence correctly, you have to know something beyond the visible structure of the model sentences. This point is the critical one in the poverty of the stimulus argument: the rules of English and any other natural language are too “rich” to be discovered simply by studying the speech an individual encounters while learning the language.

The result of Chomsky’s revolution is apparent around us. Behaviorism ceased to be America’s dominant school of psychology. It has been replaced by a cognitive psychology that tries to explain all human behavior in terms of input, computation, and output. Chomsky’s work itself gained enormous prestige as the weapon of a giant killer and came to hold a central position in modern linguistics.

As it happens, I was never very enthusiastic about cognitivism either, because it still does not explain the literary and artistic achievements of humanity that have always been my central interests. I also quickly saw that Chomskyan linguistics too was not going to help me with those matters, and I did not pursue it. But I have always appreciated Chomsky’s role in putting a halt to Skinner’s attempt to explain literally everything in terms of the environment.

Thus, I was surprised when a commenter began insisting on the importance of operant conditioning in learning language, and I challenged him to provide a rebuttal to the poverty of the stimulus argument. To my delight, the fellow posted a long bibliography. The argument goes on.

This introductory post is already long enough, so I will put up the remainder of the analysis on following two days to consider just where we now stand and whether the idea that speech has an evolutionary history (in the biological sense) can still stand. After I’m done, the commenter will be a guest blogger and make his response.

Comments

[JanetK]

I notice an implication (maybe only in my perception but maybe not) that Chomsky in the late 50s was the person who first weakened Skinnerism. Was the birth of cognitive psychology not in the psychologists that helped the Air Force in WW2 to train pilots in shorter and shorter times? I have no reference for this, I just remember encountering it before I encountered Chomsky.
---------------------
BLOGGER: Cognitive psychology is much older than Chomsky's review of Skinner but it was very much in second (or maybe third) place behind behaviorism (aka, learning theory).

[isakerem]

that is a very important post. I believe that Chomskian Revolution is some kind of quantum of linguistics. That is, we were so used to language that we thought everything could be explained with simple mechanisms without much effort. Yet, by amazing us with the complexity and unknown aspects of language, chomsky created a new paradigm to be studied on. I believe that Chomsky's distinction between competence and performance is also a challenge to behaviorism because it shows that it is not possible to analyze language only with "observable" behaviour. Production is observable but competence is not. So the question is this: How can a child of seven reach the competence of a complex phenomenon like language while he cannot even do basic calculations in maths? By this I mean that language and general intelligence are not directly related, which is also a challenge to behaviourism because they would like to explain language as a result of brain development. However, the case with people with Williams syndrome is interesting ( and that is why I write in the last post against the idea that crows' tool use may be an indication of sth about language) With an IQ of 50, they can form perfectly grammatical and semantically complex sentences. Language seems to some kind of instinct (Pinker) Lastly, I want to say that imitation and correction are not the mechanisms under the language development because firstly parents seem to make no correction of syntax but only and rarely of semantics and secondly children can form brandnew sentences in a very short time (which shows that imitation lacks the explanatory pover).

[mariana]

Chompsky said:you cannot determine the rules for organizing sentences simply by studying the physical organization of sentences. This new idea changed the thought that there could be a rigid structure that modelizes language, and that that structure could be infered easily from listening to language.
I guess this is related to he birth of NLP. Where we now have several learning methods that help us aquire the language model at several levels (syntactic, morphologic,..). And this nlp methods are still incomplete to model the hole of the language, they have several failures that still need to be addressed somehow.