What You Can Change _. And What You Can't - Martin E. Seligman [36]
Maybe they had just failed to notice the bright noise during conditioning? So Garcia counterbalanced the experiment. Other rats were given the same bright, noisy saccharin paired now with foot shock instead of stomach illness. What did they learn? They now cringed in fear of bright noise, but they still loved saccharin. When they suffered pain, they blamed it on the bright noise and ignored the taste.
So both the bright noise and the saccharin were noticeable, but only the taste became aversive when the rats became nauseated, and only the bright noise became aversive when the rats suffered pain. How could this be? Both the Garcia findings and the sauce béarnaise aversion looked like Pavlovian conditioning. But they did not fit the laws of conditioning. There were five problems:3
First, Pavlovian conditioning is not selective: Any CS that is present when any UR occurs should get conditioned. Pavlov was there when the dogs were fed, and so he became exciting (so, presumably, did his voice, his after-shave lotion, and his white coat). But this didn’t work for Garcia’s rats. Only the saccharin became aversive with stomach illness, and only the bright noise became aversive with pain.
Similarly with sauce béarnaise; it had selectively absorbed the badness of the incident—at the expense of other potential CSs. The sauce béarnaise, and nothing else, became nauseating. There were plenty of other stimuli present that should have been conditioned, since conditioning works by sheer contiguity between co-occurring events. But no other stimulus became aversive. I still liked beef. I still liked my wife, who was there the whole time. The white plates I ate the sauce off and the Danish stainless-steel silverware still looked okay to me, though they were paired with illness. So, too, was the toilet bowl I threw up into, and it didn’t look unusually noxious to me afterward, either.
The second violation of known laws was that Pavlovian conditioning bridges only very short time gaps. If you hear a burst of noise and an electric shock follows it by one second, you will become afraid of the noise. If the shock doesn’t happen until a minute later, however, you will not become afraid of the noise. A single minute’s delay between CS and US is too long for conditioning to bridge. The delay between the saccharin and the radiation sickness was much longer, just as was the delay between tasting the sauce béarnaise and getting sick: not seconds, but hours—hundreds of times longer than in any successful conditioning experiment.
The third problem was that Garcia’s rats hated saccharin after just one dose of X rays, and I came to dislike sauce béarnaise with just a single experience. Normal Pavlovian conditioning almost never takes in just one trial. It took Pavlov repeated feeding of his dogs to get them to salivate to his presence. It takes about five noise-shock pairings to get someone sweating over noise. This is true even when the shock is very painful.
The fourth problem was that ordinary Pavlovian conditioning is rational: Its laws follow the growth and decline of conscious expectations. Pavlov’s dogs learned to expect food when Pavlov showed up, and so they salivated. Once Pavlov stopped feeding them, they stopped salivating when they saw him—this is called extinction. Sauce béarnaise didn’t work this way, however. The day after I got sick, I called my closest collaborator to apologize for not showing up at the lab. He asked me if I had come down with the stomach flu that was sweeping the department. After that I “knew” that the sauce was innocent; the stomach flu,