Why Did Psychology Give Up on Consciousness? Introspection, Fechner's Threshold and Watson's Behaviorist Manifesto, 1850–1913

This is the first part of a new series, A History of the Science of Consciousness. The sense that something is being seen or heard right now — consciousness — is visible only from the inside. How has science tried to check it from the outside? The thread running through the series is a single question: how far can we trust a person's own report of what they saw or felt? Part 1 is the starting point. In the second half of the nineteenth century psychology began as a laboratory science that measured the edges of sensation and the speed of the mind. In the early twentieth century, trained observers began to give flatly contradictory reports, and there was no way to settle between them. In 1913 John Watson proposed dropping the word "consciousness" from psychology altogether. Working from the original papers, this article asks what could be measured and what could not.

The story of a science that only looked inward

Early psychology is usually described like this. It relied on introspection — subjects looking into their own minds and reporting what they found — and so it was subjective. In 1913 the American John Watson rejected it as unscientific, and behaviorism, which studied only observable behavior, began.

That is half right. The earliest experimental psychology was mostly a set of ways to get numbers from the outside, not to have people describe their inner life. Wilhelm Wundt, later cast as the leading introspectionist, had little trust in free introspection. What broke down was something narrower: the attempt to have people report, in detail and in words, what went on in their minds while they were thinking. This article follows what could be measured, in order, and finds where the trouble started.

Measuring the threshold of consciousness: Fechner's psychophysics, 1850–1860

By his own account, the Leipzig physicist Gustav Fechner had an idea on 22 October 1850. The growth of a sensation cannot be measured directly, but the growth of the stimulus needed to produce it — light, or weight — can. Use that as the ruler. After ten years of experiments on lifted weights, brightness and distances, he published the two volumes of Elemente der Psychophysik (Elements of Psychophysics) in 1860.

His starting point was a finding by Ernst Weber, a physiologist in the same city: the smallest difference in weight that a person can notice grows in proportion to the weight itself. Fechner reasoned that if you counted these "just noticeable differences" one by one, you could put a scale on sensation. The result was his law: the strength of a sensation goes with the logarithm of the strength of the stimulus. Put another way, when the stimulus grows by the same factor, the sensation grows by the same step. In Fechner's own arithmetic example, adding 10 to 10 raises the logarithm by a third, while adding 10 to 1,000 raises it by about one seven-hundredth. One candle stands out in a dark room and goes unnoticed in a bright one.

The law has a second pillar. A stimulus that is too weak is not felt at all. That boundary is the threshold. Fechner gave the sensations from stimuli below threshold negative values in his formula and called them unconscious sensations; sensations above threshold he called conscious. The first quantity psychology ever had in hand was the boundary between what is conscious and what is not.

The subjects in these experiments reported very little: whether they felt something or not, whether they could tell two things apart or not. The same trial was repeated many times and the answers counted. Nobody had to describe the content of anything. Fechner also expected that the relation between sensation and activity in the brain — what he called inner psychophysics — would follow the same kind of law, but with what was then known about the brain this remained an expectation that could not be tested.

Measuring the speed of the mind: Donders's subtraction, 1865–1868

Time became measurable as well. In 1850 Hermann von Helmholtz estimated the speed of a signal along a frog's nerve at about 30 metres per second. Nerve signals take a measurable time. Prompted by his friend's work, the Utrecht physiologist Franciscus Donders set out to measure the time taken by thinking itself.

According to his laboratory notebook, the decisive experiment was run on 21 August 1865. One person spoke a syllable and another repeated it as fast as possible. Both voices were traced as waves on soot-covered paper wrapped round a turning drum, and the time was found by counting the vibrations of a tuning fork recorded alongside. There were three methods. In the a-method the syllable was always "ki", and the listener repeated it. In the b-method the speaker said any of "ka", "ke", "ki", "ko" or "ku", and the listener repeated whatever he heard. In the c-method the speaker said any of the five, and the listener answered only to "ki" and stayed silent otherwise.

With Donders himself answering, the a-method took 201 milliseconds, the c-method 237 and the b-method 284. Donders reasoned by subtraction. The c-method is the a-method plus the work of telling which sound it was; the b-method adds the work of choosing which sound to say. So telling the sounds apart took 36 milliseconds, and choosing took 47. He published the result in 1868.

Here too the subject said nothing about his inner life. He only spoke a syllable. Yet the steps inside the mind had numbers attached to them.

But what the numbers mean rests on an assumption about what happens inside. In 1874 Wundt pointed out that the c-method also involves a choice — whether to answer or to stay silent. If so, the gap between b and c is not the whole time spent choosing. Donders himself wrote that many people found the c-method hard: they answered when they should have stayed silent, and a single slip meant the whole series had to be thrown out. For that reason he put most weight on his own records, which were free of errors. In 2018 the psychologist Ardi Roelofs went back to Donders's notebook and found that for two of his students, measured the same day, the c- and b-methods took almost the same time. When Roelofs repeated the experiment on himself with Donders's original lists of syllables, the order a, then c, then b came out again. The clock reads the same for everyone. What the subtraction stands for may have differed from person to person.

Introspection inside the laboratory: Wundt and James, 1879–1896

In 1879 Wundt began a psychology laboratory at the University of Leipzig, often taken as the starting point of experimental psychology. His introspection did not mean looking freely into one's mind. In his 1896 textbook Grundriss der Psychologie (Outlines of Psychology) he explained why. Mental events are not fixed objects but processes in constant change, and the very intention to observe them alters them or makes them disappear. Exact observation is therefore possible only as experiment, where the experimenter decides the moment the process begins and can vary the conditions and repeat it. What Wundt studied this way were simple processes: sensations, simple judgments, reactions. Higher mental life — language, myth, custom — he said should be studied not by experiment but by observing the products people leave behind.

In America, William James wrote in his Principles of Psychology (1890) that introspection is what we must rely on first, foremost and always. In the same chapter he admitted that introspection is difficult and fallible, just as all observation is. The mind that says "I feel tired", he added, is no longer simply tired; it is in the state of saying that it is tired.

James described consciousness as an unbroken stream. The hardest parts to see from the inside are the passages from one thought to the next: stop to look at them and they are no longer passages. Trying, he wrote, is like grabbing a spinning top to catch its motion, or turning up the gas quickly enough to see what the darkness looks like.

Imageless thought: Würzburg against Cornell, 1901–1909

In the early twentieth century psychologists at Oswald Külpe's institute in Würzburg began to bring thinking itself into the laboratory. They set a task, then had the observer report in as much detail as possible what had happened in his mind meanwhile. In word-association experiments published by Mayer and Orth in 1901, observers kept reporting "states of consciousness" that were neither images nor feelings.

In 1907 Karl Bühler went further. He asked questions that could not be answered without thinking, to be answered "yes" or "no"; he timed the answer with a stopwatch; then he had the observer describe the experience in full. The questions were of the kind "Was the Pythagorean theorem known in the Middle Ages?" and "Can our thought grasp the nature of thought?" Sometimes the observer read an aphorism of Nietzsche and was asked whether he understood it or agreed. Bühler relied on two experienced observers, the institute's director Külpe and Dürr. Asked whether thought can grasp the nature of thought, Külpe answered "yes" after six seconds. Hegel's objection to Kant had come to him suddenly, he said; he knew exactly what it was about, but there were no words in it and no images. The word "Hegel" turned up only afterwards. Bühler concluded that thoughts as such — neither words nor images — are present in consciousness.

Wundt attacked this the same year. Such questioning experiments, he wrote, only look methodical because they happen in a psychology laboratory with two people playing experimenter and observer. They are not experiments at all, and have none of the features that set the introspection of experimental psychology apart from everyday self-reflection. As the previous section showed, for Wundt introspection counted as observation only when the moment of onset was fixed and the conditions could be varied and repeated.

The challenge also came from the other side. At Cornell University in America, Edward Titchener opened his 1909 lectures on thought by describing his own mind. It was full of images. He never sat down to read without music playing in his head. Even "meaning", the most abstract of things, appeared to him as a picture: the blue-grey tip of a kind of scoop digging into a dark, plastic mass. When he asked the twelve students in his graduate seminar, two at once said they had similar pictures. Titchener openly admitted that his mind leaned toward images, and added that minds without images must lean just as far the other way.

This is the heart of the problem. Trained observers on both sides were reporting honestly. Yet on the most basic point — whether a thinking mind contains images — their answers were opposite. And there was no measure outside the reports themselves to decide who was right.

Taking out the word "consciousness": Watson's manifesto, 1913

In 1913 John Watson, who studied animal behavior at Johns Hopkins University, published "Psychology as the Behaviorist Views It" in the Psychological Review. It opened by declaring psychology a purely objective, experimental branch of natural science whose goal is the prediction and control of behavior. Introspection, he wrote, forms no essential part of its methods.

His reason was exactly the dispute of the previous section. In physics or chemistry, when an experiment fails to replicate, you suspect the apparatus or the chemicals. In psychology, Watson complained, if you could not reproduce my findings, the fault was your untrained introspection. In a footnote he named the controversy over imageless thought. Unless introspection was abandoned, he predicted, psychology two hundred years on would still be divided over whether auditory sensations have the quality of "extension". In its place he proposed to put humans and animals on the same plane and build psychology in terms of stimulus and response and the forming of habits.

Watson did not throw out human speech, however. Having a subject adjust an apparatus until two lights look the same, or say whether a stimulus is there or not, he accepted as a "language method" — an abbreviated way of studying behavior. A report was to be treated not as a window onto the mind but as one more piece of behavior. He also admitted that complex processes such as imagination, judgment and reasoning could not yet be studied by behavior methods, and suggested they be set aside for a time. He added his expectation that thinking would turn out to be faint movements of the muscles of the larynx.

Near the end of the paper Watson asked only this: that behaviorists be allowed to use consciousness the way other natural scientists do, without making it a special object of observation. That, he said, was all his thesis required. Behaviorism went on to become a major current in American psychology.

What could be done at this point, and what could not

What could be measured were thresholds of sensation and the time taken to respond. Both asked the subject for the barest report — felt or not, different or not, speak or stay silent — repeated many times and turned into proportions and averages. Fechner's threshold showed that the very boundary between conscious and unconscious could be handled as a number.

What could not be measured was the content of experience. Does a thinking mind contain images? When trained observers disagreed, there was no measure to decide between them. Even the measurements that needed the barest reports, as Donders's subtraction shows, rested on assumptions about the steps inside the mind. Watson's answer was to stop treating reports as windows onto the mind and treat them as behavior. That made the disagreement stop mattering; it did not resolve it. The question of what experience is like was set aside, untouched.

Part 2 turns to ways of measuring consciousness without asking the person anything. In 1924 the German psychiatrist Hans Berger recorded the brain's electrical rhythms through the human scalp. Awake or asleep: not the content of consciousness but its level began to be measurable without any report, in the years 1924–1953.