by DK
"The test of a psychological theory, as well as its moral justification, lies in its application to concrete practical considerations."
Donald Broadbent
Broadbent was interested in using psychology not only to design better equipment, but also to reach a better understanding of what affected the pilots’ capabilities. They clearly had to cope with large amounts of incoming information, and then had to select the relevant data they needed to make good decisions. It seemed to him that mistakes were often made when there were too many sources of incoming information.
Broadbent was influenced in his thinking about how we process information by another product of wartime research: the development of computers and the idea of “artificial intelligence.” The first director of the APU, Kenneth Craik, had left the unit important manuscripts and flow diagrams comparing human and artificial information processing, which Broadbent clearly studied.
"His psychology was intended for society and its problems, not merely for the dwellers in ivory towers."
Fergus Craik and Alan Baddely
At the same time, code breakers such as the mathematician Alan Turing had been tackling the notion of information processing, and in the postwar period he applied this to the idea of a “thinking machine.” The comparison of a machine to the workings of the brain was a powerful analogy, but it was Broadbent who turned the idea around, considering the human brain as a kind of information-processing machine. This, in essence, is what distinguishes cognitive psychology from behaviorism: it is the study of mental processes, rather than their manifestation in behavior.
To study how our attention works, Broadbent needed to design experiments that would back up his hunches. His background in engineering meant that he would not be satisfied until he had evidence on which to base a theory, and he also wanted that research to have a practical application. The APU was dedicated to applied psychology, which for Broadbent referred not only to therapeutic applications, but also to applications that benefited society as a whole; he was always very conscious that his research was publicly funded.
One voice at a time
One of Broadbent’s most important experiments was suggested by his experience with air traffic control. Ground crew often had to deal with several streams of incoming information simultaneously, sent from planes arriving and departing, which was relayed to the operators by radio and received through headphones. The air traffic controllers then had to make quick decisions based on that information, and Broadbent had noticed that they could only effectively deal with one message at a time. What interested him was the mental process that must take place in order for them to select the most important message from the various sources of incoming information. He felt that there must be some kind of mechanism in the brain that processes the information and makes that selection.
The experiment that Broadbent devised, now known as the dichotic listening experiment, was one of the first in the field of selective attention—the process our brains use to “filter out” the irrelevant information from the masses of data we receive through our senses all the time. Following the air traffic control model, he chose to present aural (sound-based) information through headphones to the subjects of his experiment. The system was set up so that he could relay two different streams of information at the same time—one to the left ear and one to the right—and then test the subjects on their retention of that information.
As Broadbent had suspected, the subjects were unable to reproduce all the information from both channels of input. His feeling that we can only listen to one voice at once had been confirmed, but still the question remained as to exactly how the subject had chosen to retain some of the incoming information and effectively disregard the rest.
Thinking back to his initial training as an engineer, Broadbent suggested a mechanical model to explain what he felt was happening in the brain. He believed that when there are multiple sources of input, they may reach a “bottleneck” if the brain is unable to continue to process all the incoming information; at this point, there must be some kind of “filter” that lets through only one channel of input. The analogy he uses to explain this is typically practical: he describes a Y-shaped tube, into which two flows of ping pong balls are channeled. At the junction of the two branches of the tube, there is a flap that acts to block one flow of balls or the other; this allows balls from the unblocked channel into the stem of the tube.
A question still remained, however: at what stage does this filter come into operation? In a series of experiments that were variations on his original dichotic listening tasks, Broadbent established that information is received by the senses and then passed on in its entirety to some kind of store, which he called the short-term memory store. It is at this stage, he believes, that the filtering occurs. His description of how and when information is selected for attention is known as the “Broadbent Filter Model,” and it demonstrated a completely new approach to experimental psychology, not only in combining the theoretical with the practical, but also in considering the workings of the brain as a form of information processing.
Air traffic controllers have to deal with a multitude of simultaneous signals. By re-creating this problem in listening experiments, Broadbent was able to identify attention processes.
The cocktail party problem
Broadbent was not the only person to address the problem of selective attention. Another British scientist, Colin Cherry, also investigated the subject during the 1950s. Working in communication rather than psychology, Cherry posed what he called the “cocktail party problem:” how, at a party where lots of people are talking, do we select which of many conversations to give our attention to, and which to ignore? And how is it possible to be distracted from our focused attention on conversation “A” by conversations “B” or “C”?
To help answer these questions, Broadbent turned his attention to the nature of the filter in his model. Precisely what information does it filter out, and what does it allow through? Following another process of rigorous experimenting, he found that the selection is made not on the content of the information (what is being said), but on the physical characteristics of any message, such as clarity or tone of voice. This suggests that even though information is stored, albeit very briefly, in short-term memory, it is only after filtering that it is processed for meaning and actually understood. This finding had important implications when applied to air traffic control, for example, where decisions could be made on possibly irrelevant or inaccurate information, rather than being prioritized according to meaning and importance.
Broadbent and Cherry worked together on many dichotic listening experiments to test the filtering process. They realized that filtering is also affected by expectation. In one experiment, participants were asked to listen to different sets of numbers presented simultaneously to each ear. In some cases they were instructed which ear (the information channel) they would be asked about first; in others no instructions were given. The results showed that when people know which ear is receiving the stream of information they will be asked for first, they switch attention to that ear, and the information that enters the other ear is not always accurately retrieved from memory. In all cases the information that people chose or were asked to remember first seemed to be processed more accurately than the later material; it was thought this might be due to parts of the information being lost from the short-term memory store before the participant tried to retrieve it. In 1957, Broadbent wrote: “We can listen to only one voice at once, and the first words we hear are the be
st recalled.”
People at a cocktail party may be listening to one conversation, but then become aware of (and switch attention to) another, if it includes personally significant information.
Modifying the model
In 1958, Broadbent published the results of his research in a book, Perception and Communication, which effectively outlined a framework for studying attention, comprehension, and memory. The timing was significant, as it coincided with a divergence of opinion about the importance of behaviorism in the US, and the book slowly became known as one of the landmarks in the development of the new cognitive psychology. As a result, Broadbent was recognized, by his peers if not the public, as the first major psychologist Britain had produced, and was rewarded the same year by being appointed director of the APU to succeed Bartlett.
Not one to rest on his laurels, however, Broadbent saw his new appointment as an opportunity to continue his work on attention, widening the scope of his research and refining the theory. From the starting point of his filter model, he returned to the cocktail party problem, and in particular one phenomenon Cherry had identified concerning the nature of information that is selected for attention. When an overheard conversation includes information that has some kind of special significance for a person—such as a personal name—the attention is switched toward that conversation, and away from the one previously attended to.
Further dichotic listening experiments at the APU bore out Cherry’s findings: attention is filtered by physical characteristics but also by meaning, using feedback from memory stores, prior experience, and expectations. The sound of a siren, for instance, would divert attention on to that stream of sound. This suggests that information is in some way understood before being selected for attention.
Broadbent realized that his filter model needed modification, but was pleased rather than dismayed to have to make the changes. As a scientist, he felt that all scientific theories are temporary, derived from the evidence available at the time, and so susceptible to change in the light of new evidence; this is how science progresses.
The work of the APU centerd around Broadbent’s research into attention, but this allowed for a constantly widening range of applications. Broadbent worked tirelessly to ensure that his work was practically useful, examining the effects of noise, heat, and stress on attention in work environments, and he constantly reviewed his ideas as he worked. In the process, he gained government support for his ideas, and the respect of many industries whose practices were improved by his work. This led to yet more research into areas such as differences of attention between individuals, and lapses of attention and their causes. In each case the results of his experiments led to refinements of his theories. In 1971, he published a second book, Decision and Stress, which detailed an extended version of his filter theory. Like its predecessor, this book became a classic textbook of cognitive psychology.
Complex industrial processes could be transformed in efficiency, Broadbent thought, through the application of psychology. He was committed to producing genuinely useful research.
The cognitive approach
Broadbent’s books did not reach the general public, but were widely read by scientists from other disciplines. His comparison of the workings of the human brain with electronic machines became more and more relevant as interest in computing increased. His model of the various stages of human information processing—acquisition, storage, retrieval, and use—echoed the work on artificial intelligence at that time.
Broadbent was instrumental in setting up a Joint Council Initiative on Cognitive Science and Human-Computer Interaction, which helped shape the development of cognitive science. His work also established applied psychology as an important approach for problem-solving, increasing its impact well beyond the confines of the laboratory. A key figure in the founding of cognitive psychology, his research into attention laid the groundwork for a new field of enquiry that continues to yield rich results today.
DONALD BROADBENT
Born in Birmingham, England, Donald Broadbent considered himself to be Welsh, since he spent his teenage years in Wales after his parents’ divorce. He won a scholarship to the prestigious Winchester College, then joined the Royal Air Force aged 17, where he trained as a pilot and studied aeronautical engineering.
After leaving the RAF in 1947, he studied psychology under Frederic Bartlett at Cambridge, then joined the newly founded Applied Psychology Unit (APU), becoming its director in 1958. Married twice, he was a shy, famously generous man whose “puritanical streak” led him to believe that his work was a privilege and should always be of real use. In 1974, he was awarded the CBE and appointed a fellow of Wolfson College, Oxford, where he remained until his retirement in 1991. He died two years later of a heart attack, aged 66.
Key works
1958 Perception and Communication
1971 Decision and Stress
1993 The Simulation of Human Intelligence
See also: René Descartes • George Armitage Miller • Daniel Schacter • Frederic Bartlett
IN CONTEXT
APPROACH
Memory studies
BEFORE
1878 Hermann Ebbinghaus conducts the first scientific study of human memory.
1927 Bluma Zeigarnik describes how interrupted tasks are better remembered than uninterrupted ones.
1960s Jerome Bruner stresses the importance of organization and categorization in the learning process.
AFTER
1979 Elizabeth Loftus looks at distortions of memory in her book Eyewitness Testimony.
1981 Gordon H. Bower makes the link between events and emotions in memory.
2001 Daniel Schacter publishes The Seven Sins of Memory: How the Mind Forgets and Remembers.
Memory was one of the first fields of study for psychologists in the 19th century, as it was closely connected with the concept of consciousness, which had formed the bridge between philosophy and psychology. Hermann Ebbinghaus in particular devoted much of his research to the scientific study of memory and learning, but the next generation of psychologists turned their attention to a behaviorist study of learning, and “conditioning” replaced memory as the focus of research. Apart from a few isolated studies, notably by Bluma Zeigarnik and Frederic Bartlett in the 1920s and 30s, memory was largely ignored as a topic until the “cognitive revolution” took place following World War II. Cognitive psychologists began to explore the idea of the brain as an information processor, and this provided a model for the storage of memory: it was seen as a process, whereby some items passed from short-term or working memory into long-term memory.
By the time Endel Tulving finished his doctorate in 1957, memory was once more a central area of study. Forced to abandon the study of visual perception due to a lack of facilities, Tulving turned his attention to memory. The funding deficit also shaped his approach to the subject, designing experiments that used no more than a pen, some paper, and a supply of index cards.
"Relating what we know about the behavior of memory to the underlying neural structures is not at all obvious. That’s real science."
Endel Tulving
The free-recall method
Learning about the subject as he went along, Tulving worked in a rather unorthodox way, which occasionally earned him criticism from his peers, and was to make publishing his results difficult. His maverick instincts did, however, lead to some truly innovative research. One hurriedly designed, ad hoc demonstration to a class of students in the early 1960s was to provide him with the model for many later experiments. He read out a rando
m list of 20 everyday words to the students, and then asked them to write down as many as they could recall, in any order. As he expected, most of them managed to remember around half of the list. He then asked them about the words that they had not remembered, giving hints such as “Wasn’t there a color on the list?,” after which the student could often provide the correct answer.
Tulving developed a series of experiments on this “free recall” method, during which he noticed that people tend to group words together into meaningful categories; the better they organize the information, the better they are able to remember it. His subjects were also able to recall a word when given a cue in the form of the category (such as “animals”) in which they had mentally filed that word. Tulving concluded that although all the words memorized from the list were actually available for remembering, the ones that were organized by subject were more readily accessible to memory, especially when the appropriate cue was given.
In Tulving’s free recall experiments, people were asked to remember as many words as possible from a random list. “Forgotten” words were often recalled using category cues. They were stored in memory but temporarily inaccessible.