I majored in psychology at Reed College. At the time, the whole major centered on Skinnerian psychology — the importance of reward in controlling behavior. The introductory course used a Skinnerian textbook (e.g., we learned the correct meaning of “negative reinforcement” — it does not mean punishment). Other courses also had a Skinnerian emphasis. They never convinced me. I always thought it was an exceedingly narrow way to study behavior.

When I was a graduate student, I visited Harvard and heard Skinner give a talk, titled “Why I am not a cognitive psychologist”. During the question period I asked if he was familiar with the work of Saul Sternberg — perhaps the most influential cognitive psychologist. No, said Skinner. I thought it was foolish to criticize an area of research you know little about.

After I became a professor, I went back to Reed to give a talk. After the talk, I went out to dinner with several psychology professors. I told them I thought Skinner was a brilliant engineer — the Skinner box is really useful — but a mediocre scientist. He was unable to discover anything, he just repeated the same result (rewarding something increases how often it is done) countless times. They had no reply.

In the last two days, strangely enough, Skinner has come up in two different conversations. In the first, a friend said that Skinner’s views about language were ridiculous. I agreed. Why write such nonsense? my friend asked/complained. I said maybe Skinner’s productivity system worked too well. It caused him to write when he had nothing to say. In the second, a different friend brought up David Freedman’s recent Atlantic article called “The Perfected Self”, which argues that Skinnerian techniques really work when you implement them as smartphone apps — techniques to lose weight, for example. “B. F. Skinner’s notorious theory of behavior modification was denounced by critics 50 years ago as a fascist, manipulative vehicle for government control,” writes Freedman (or an editor), but actually that theory is really good.

My area of academic psychology (animal learning) is the same as Skinner’s. Within this field, I have never heard anyone complain that Skinner’s work was “fascist” or “manipulative” or a “vehicle for government control.” It never became popular — it was always a minority point of view — probably because it was boring (the same thing over and over) and perhaps because it was anti-intellectual. Skinner wrote a well-known paper about why theories are unnecessary. He didn’t understand the role of theories in science and didn’t bother to find out. Sure, the psychology theories of the time (1950) were awful. Psychology theories are still mostly awful. But there are plenty of good theories in other areas of science.

For a long time, Skinnerian ideas, nearly dead in academia, lived on in the treatment of autism. The people applying these ideas called themselves “behavior analysts” and the whole field of applied Skinnerian psychology was called “behavior analysis”. What caused this persistence was that the techniques worked. Using the techniques (carefully rewarding this or that behavior) improved the lives of autistic children and their parents. Which was a real contribution. I could make a long list of famous psychologists who have done less to improve human well-being.

The success of Skinnerian ideas in improving the lives of autistic children should not be confused with figuring out what causes autism. To figure out the cause of autism is to figure out the environmental cause(s) — to which people with certain genes are more sensitive — and how autism can be avoided entirely, not meliorated. I have blogged about possible causes of autism many times, in particular the possibility that sonograms cause autism. I have no idea if behavior analysts understand the difference between melioration and figuring out the cause. Maybe Skinner would claim there is no difference — he was full of bizarre statements like that. If your child is autistic, you are in crisis. You have zero interest in questions about “cause” — you simply want help. In any form. Behavior analysts, while helping autistic children and their parents, contribute nothing that helps us find the cause of autism. Which, if you are planning on having children, you care about enormously. So you can avoid having autistic children.

So Skinner’s legacy is mixed. The Skinner box is terrific. I happily used them in my research for years, even though I hardly believed a single word Skinner said. As an engineer — an applier of stuff discovered by others — Skinner made a lasting contribution. As a self-promoter, he was incredibly successful — he was on the cover of Time, for example. As a scientist, he was a zero. He discovered nothing that matters. As a thinker (e.g., the book Beyond Freedom and Dignity) he was less than zero. He was a charlatan, claiming over and over that he understood puzzling things (e.g., language) that he did not understand. An unusual mix. Few great engineers are charlatans.

In a great essay, Edward Jay Epstein points out, at least by implication, that the Pulitzer Prize committee is not terribly interested in the truth of things:

A sustaining myth of journalism holds that every great government scandal is revealed through the work of enterprising reporters who by one means or another pierce the official veil of secrecy. . . This view of journalistic revelation is propagated by the press even in cases where journalists have had palpably little to do with the discovery of corruption. Pulitzer Prizes were thus awarded this year to the Wall Street journal for “revealing” the scandal which forced Vice President Agnew to resign and to the Washington Star/News for “revealing” the campaign contribution that led to the indictments of former cabinet officers Maurice Starts and John N. Mitchell, although reporters at neither newspaper in actual fact had anything to do with uncovering the scandals. . . . Yet to perpetuate the myth that the members of the press were the prime movers in such great events as the conviction of a Vice President and the indictment of two former cabinet officers, the Pulitzer Prize committee simply chose the news stories nearest to these events and awarded them its honors.

The Nobel Prize in Biology committee operates the same way, except with the disadvantage that there is not one important (= useful in a big way) biology discovery per year. There are far fewer than that. So almost every year the Nobel Prize in Biology goes to discoveries with little practical importance that are described as having great practical importance. The profession (in this case, biology) is credited with much more power than it actually has.

Why does this happen? One possible reason is that no one points it out. (Epstein’s essay, still relevant today, was published in 1974.) When a powerful journalistic institution does bad things, it is incredibly dangerous (to your career) to point this out. This is why the Murdoch scandal is so big — it went on so long. Spy magazine had a column called Review of Reviewers. It was hilarious because the misdeeds were great. Unlike almost anything else in Spy, the author was anonymous. Brilliant writing that the author did not take credit for because it was dangerous to criticize the watchdogs. Likewise, hardly anyone except Epstein criticizes the prize committees (who resemble watchdogs) so they can be profoundly inaccurate.

Drug companies, in a few cases, have recently started to release much more data from drug trials. Unsurprisingly, analysis of the new data by outsiders — people who have nothing to gain from positive results — has often contradicted the drug company analysis of the same data.

One example involves the flu drug Tamiflu. The new analysis suggested that “Tamiflu falls short of claims—not just that it ameliorates flu complications, but also that the drug reduces the transmission of influenza.” Another example involved Prozac. The new analysis “ended up bucking much of the published literature on antidepressants. . . . [It]found no link between Prozac and suicide risk among children and young adults . . . Prozac appeared to be more effective in youth, and antidepressants far less efficacious in the elderly, than previously thought.”

Another reason to believe in the value of this new data is the work of Lisa Bero at UCSF. She looked at the efficacy of nine drugs using unpublished FDA data. “Nineteen of the redone analyses showed a drug to be more efficacious, while 19 found a drug to be less efficacious. The one harm analysis that was reanalyzed showed more harm from the drug than had been reported.”

I hope that the FDA will eventually require that all raw data from drug trials be publicly available as a condition of approval. (The same should also be true of journal articles, as a condition of publication.) It is abundantly clear that drug company analyses are often misleading — which harms the public.