As the so-called cognitive revolution took shape, a sense of urgency arose to dispatch behavioristic views. It’s said that every intellectual movement has a dragon, and must declare it slain to justify its existence. For early Behaviorism, the dragon was Introspectionism. For the cognitive revolution, the dragon was behaviorism.
As has been detailed many places, the early cognitive revolution (circa 1940-1950) was inspired, in part, by the notion that “thinking machines” (computers) provide a model of how human thinking works. Since computers “process information,” this approach necessarily assumed the same of humans. A key sticking point was the behaviorist research program demonstrating considerably sophisticated learning, mainly in nonhumans, as a result of no apparent “information” except experience with operant contingencies.
Since information was the presumed currency of the mind, cognitivists posited that humans, when exposed to contingencies, must be generating their own “information.” That is to say, they must become aware of the contingencies, a view that quickly transformed into the hypothesis that (unlike nonhumans, apparently), humans could only learn from contingencies if they were consciously aware of them.
This hypothesis was notoriously difficult to test. For instance, in Greenspoon’s famous experiment, participants were asked to talk, and a listener-experimenter would make noises of approval (“Mmm-hmm”) following plural nouns, which subsequently increased in frequency, apparently showing reinforcement effects. When interviewed, the participants seemed unaware of this effect or the contingency that spawned it. But arguments arose as to whether awareness was adequately probed, and some replication efforts, purporting to fully eliminate the potential for awareness, did not reproduce the results. The proposed conclusion: “There is no convincing evidence for operant or classical conditioning in adult humans.”
Some people might also point to Paul Fuller’s (1949) famous early demonstration of shaping in a comatose patient, but with definitions of “coma” pretty hazy, and uncertainties about how much residual awareness might linger in an unresponsive individual, it was easy to dismiss this study as not necessarily showing what it claimed to.
Image credit: Alchetron
But lurking in the background was research that demonstrated “learning without awareness” beyond any reasonable doubt. This now mostly-forgotten research was conducted by Ralph Hefferline, and it employed a brilliant setup in which only the stubbornest of critics could argue for some sort of awareness.
Hefferline was an interesting guy. He received his PhD from Columbia in 1944 (where he did some operant research), and went on to join and later chair the Psychology faculty at Columbia. He was better known as a Gestaltist than a behavior analyst, and co-authored an well-regarded 1951 book on Gestalt therapy. But he was also much influenced by Skinner’s accounts of private events and verbal behavior (we know about the latter because Hefferline took notes on a 1947 summer course at Columbia in which Skinner presented early versions of what would become Verbal Behavior; distribution of these notes helped to spread Skinner’s ideas in advance of the book’s publication).
Hefferline also had some exposure to biofeedback methods which, when combined with an interest in private events, provided the perfect vehicle for researching the role of awareness in operant learning. Hefferline published a series of studies employing similar methods; here to illustrate I’ll focus on his 1959 report in Science.
Participants were fitted with headphones and several electrodes, one of which could measure extremely tiny movements in a thumb muscle. Noise was played through the headphones, but would temporarily cease when a tiny thumb twitch occurred. Results for the main group are shown at right. In an A-B-A design, thumb twitches occurred in frequently in baseline but increased in rate under the avoidance contingency. Rates decreased when the contingency was suspended.
When queried afterward, these participants claimed to have done nothing to stop the noise. Because they were unaware of making thumb twitches, they could not know that thumb twitches terminated the noise. Learning without awareness.
But maybe the coolest thing about this experiment was the control groups. One was told that some unspecified, unobservable response could terminate the noise. One was told that the critical response was a tiny thumb twitch, and another was told the same and given visual feedback to show when the response occurred. All four groups performed similarly, effectively showing that “awareness” added nothing to the effect of the avoidance contingency.
It would be awesome to say that Hefferline’s research bowled over the cognitive critics, pitching the whole awareness-as-prerequisite hypothesis out the window. but that hypothesis hung on stubbornly. More than two decades later Brewer (1974) still asserted
Conditioning in human subjects is produced through the operation of higher mental processes, rather than vice versa.
Oy. But gradually cognitive psychology had to come to grips with mountains of evidence that not everything results from conscious consideration of “information.” Today it’s widely accepted, even in cognitive psychology, that a lot of behavior control traces to processes that are variously described as implicit, unconscious, automatic, etc. (in other words, outside of conscious awareness). Hefferline’s work stands as a hearty “I told you so.”