Authors: Frank Tallis
In an important series of studies conducted principally by Norman Dixon, of University College London, it was demonstrated that the visual threshold of one eye could be altered by the subliminal presentation of emotional or neutral words to the other. The visual threshold was assessed by gradually increasing the brightness of a spot of light from levels too faint for visual recognition to a level at which subjects could determine its first appearance. Dixon discovered that more intense light was required to ensure visibility of the spot in one eye when subjects were receiving subliminal presentations of words such as whore and cancer (rather than emotionally neutral words) to the other eye. Even though the taboo words did not enter awareness, their presence seemed to be producing a general elevation of the perceptual threshold. In other words, a defensive response. This kind of experiment, which did not require subjects to say taboo words aloud, was beyond the criticism aimed at earlier perceptual defence studies.
How is it that the perceiver can selectively defend against an emotional stimulus unless it has already been identified? Perceptual defence seems to depend on the offices of an unconscious censor, who evaluates all incoming information before making a judgement concerning its suitability for entry into awareness. Perceptual defence is, of course, merely a specific instance of what must be happening all the time if consciousness is a limited-capacity channel.
Many years after conducting the first perceptual defence studies Jerome Bruner colourfully suggested the conscious mind must be equipped with something like a ‘Judas eye’ – the peephole used by bouncers at speakeasies to distinguish between members (for whom the door would open) and the police (to whom the door would remain closed).
But in a perceptual defence task, who is looking through the Judas eye? Who is opening the door – or at least leaning on it in an attempt to keep it closed?
It seemed unlikely to post-war psychologists that the mind harboured some kind of unconscious secondary personality (or superego), permanently engaged in the task of editing experience before it was deemed suitable for entry into awareness. Subsequently psychologists looked to the new computer-based model of the mind to provide them with an alternative explanation. They reasoned that if the mind functions like a computer, then it very probably processes information in a serial fashion. Perception must involve many tiers of processing, not all of which are accessible to consciousness. Indeed, only the final stages of processing might produce an event in awareness (for example, perception of a word display). Thus a stimulus exposed for a very brief period of time might be processed sufficiently to merit an adjustment in perception threshold, but insufficiently to enter awareness.
Psychologists were still discussing the operation of unconscious mechanisms, but not in the heated language of nineteenth-century Vienna. Instead, they were using the more restrained vocabulary of computer science. The new model did not require unconscious agencies or submerged secondary personalities to explain perceptual defence. All that was needed was a hierarchy of processing stages, progressing from lower to higher levels.
Shortly after the original work on perceptual defence was completed another important study was published in 1951 by R. S. Lazarus and R. A. McCleary. The experiment required subjects to be conditioned – a procedure famously developed by the Russian physiologist Ivan Pavlov. (While investigating digestion in dogs, Pavlov discovered that laboratory animals could be taught to salivate in response to a bell or light if either had previously been paired with food.) Lazarus and McCleary presented meaningless or so called nonsense syllables (always formed by placing a vowel between two consonants, as in QEX or TEG) to subjects, and paired each presentation with an electric shock. The purpose of this procedure was to condition subjects so that they would respond to any further presentations of these syllables with anxiety. Subjects were then shown the shock-associated syllables mixed with a new set of syllables that had not been paired with electric shock. A variety of exposure times were employed, using a tachistoscope. Even when the shock-associated syllables were presented for durations too brief for conscious recognition, subjects still responded by showing increased sweat-gland activity – a reliable indicator of anxiety.
Sweat-gland activity is very easily provoked, and associated with even the slightest changes in emotional state. Although the consequences of excessive sweat-gland activity become obvious in stressful circumstances – as perspiration – most people are unaware of the tiny changes that accompany more subtle emotional shifts (such as when you hear someone saying your name); however, these minute fluctuations are easily detected as changes in skin conductivity if a low current is passed between two electrodes on the skin. The machine that records changes in skin conductance is called the polygraph (although it is better known as the lie detector) and the change in skin conductance itself is known as the skin conductance response (SCR) or the galvanic skin response (GSR). Both terms have equal currency.
The fact that Lazarus and McCleary’s subjects produced large SCRs in response to subliminal presentations of shock-associated nonsense syllables was an extraordinary finding. These syllables, being nonsense syllables, had no intrinsic meaning. The only meaning they possessed had been acquired through conditioning. The brain had understood their significance (as things previously linked with pain) and responded by triggering a mild anxiety state; however, recognition and analysis had occurred in the absence of awareness. Lazarus and McCleary described the phenomenon as
subceptiort,
although, like perceptual defence, it is primarily another example of subliminal perception.
Even though phenomena such as perceptual defence and subception suggested that information could be processed outside of awareness, the experimental investigation of the brain’s preconscious processing capacity was still regarded by many as a controversial area; nevertheless, the nature of the controversy had changed somewhat. Apart from the perennial presence of those who seemed to object to anything relating to the unconscious as a matter of principle, the existence of preconscious processing was not in doubt. The real issue concerned the degree to which information could be processed in the absence of awareness.
If the brain does work like a computer, then incoming information will be analysed in a step-by-step fashion, where each step is associated with a more complex analysis. For example, the first stage might involve a very basic registration of presence (Is something there or not?); the second stage might perform a structural analysis (What does it look like?); a third stage might seek to establish emotional significance (Is it a good thing or a bad thing?); and finally, a fourth stage might execute a much more sophisticated semantic analysis (What does it mean exactly?). To what extent then, are subliminal stimuli processed? Some commentators have chosen to ask the same question using less technical language: ‘Is the unconscious smart or dumb?’
Through the observation of patients, Freud and his followers had come to the conclusion that the unconscious was very smart indeed; however, hardly any experimental work had been undertaken to prove this. With the arrival of instruments like the polygraph and the tachistoscope, psychologists were suddenly in a position to begin putting the unconscious through its paces. For advocates of a smart unconscious the early signs were promising – phenomena such as perceptual defence and subception suggested that stimuli could be analysed for meaning in the absence of awareness. But not everyone was convinced. Thus, the experimental agenda in subsequent years was very much concerned with exploring the degree to which preconscious processing could be described as superficial or complex.
In 1953 Colin Cherry published an article with the intriguing title: ‘Some experiments on the recognition of speech with one and with two ears’. He had become interested in understanding a phenomenon called the cocktail party problem, which concerns the brain’s ability to follow a conversation against the general hubbub of a party. How is it done? What features of a single human voice, among many others, does the brain lock on to? To explore the cocktail party problem, Cherry introduced a new experimental procedure called the
dichotic listening task.
Experimental subjects were given a set of headphones and asked to follow a message which was played through one channel while an entirely different message was being played through the other channel. Thus, the left and right ears were receiving separate streams of information. Cherry discovered that the ability to follow a message in one channel (while ignoring the message in the other) was largely dependent on ‘physical’ characteristics; for example, the gender of the voice or its location. When Cherry presented recordings of two different messages but spoken by the same voice – thus eliminating any superficial differences -listeners found it difficult to follow a single channel. The different messages got mixed up.
Cherry also conducted experiments in which one message was shadowed (i.e. immediately repeated back aloud) while a second message was played in the other ear. When subjects did this, very little information from the unattended message ‘got through’. For example, subjects didn’t notice when the unattended channel contained passages spoken in a foreign language. Nevertheless, so-called physical changes (such as the introduction of a pure tone) were always detected. Cherry concluded that the unattended message received little or no processing.
Cherry’s work is important for several reasons. Firstly, he more or less started systematic research into the subject of focused attention; secondly, by devising the dichotic listening task, he equipped experimental psychologists with a simple method of controlling the presentation of auditory information to subjects. Both the study of attention and the dichotic listening task would play an important role in subsequent research into the unconscious and preconscious processing.
The study of attention came to the forefront of academic psychology with the publication of Donald Broadbent’s
Perception and Communication
(1958). This was a landmark work because in it Broadbent describes a model of attention that owes much to electronic communications theory and computer technology. It assumes that the brain is best construed as an organ that processes sensory information through limited-capacity input channels. Moreover, Broadbent chose to illustrate different stages of processing with the aid of flow charts (a device already favoured by computer programmers).
Broadbent’s model was originally devised to explain how the brain copes with all of the information it receives through the senses. He suggested that the brain possesses a central information-processing channel of very limited capacity, which corresponds roughly with awareness. The central channel can only select one sensory input channel at a time, but can switch between channels at the rate of about twice every second. When we ‘pay attention’ to something, we are selecting a specific input channel for more detailed processing in the central channel. Information that is still coming in through unattended channels occupies a short term memory store for a few seconds only. Then it is lost. Broadbent called the mechanism that selects incoming information for more detailed processing the
filter.
After Broadbent, other psychologists proposed similar theories, often including a selection mechanism resembling Broadbent’s filter. Subsequently, the term ‘filter theory’ was routinely employed as a generic term.
All filter theories posit a bottleneck in the information-processing apparatus. The vast amount of incoming sensory information encounters a limited-capacity channel, which necessitates the selection of some stimuli for representation in awareness at the expense of others. If this were not the case, then human beings could simply attend to everything. The concept of ‘paying attention’ automatically implies the existence of a central processing mechanism which regulates the passage of information through the bottleneck.
Broadbent supposed that incoming information was selected for further processing at a very early stage (i.e. immediately after a superficial physical or structural analysis); therefore, Broadbent’s theory is described as an
early selection
theory. A
late selection
theory would entail a more thorough analysis (involving perhaps analysis of stimuli for meaning) before the selection was made. In either case, selection would eventually result in the admission of incoming information into awareness. Broadbent’s early selection theory provided a good framework for understanding the results that Cherry had reported after conducting his dichotic listening studies. While shadowing a message in one ear, experimental subjects could only identify superficial features of the message playing in the other ear (e.g. the gender of the speaker). Broadbent’s theory suggested that unless information entered awareness, it could only ever receive a superficial analysis. As far as Broadbent was concerned, then, the unconscious was not smart, but fairly dumb.
Within a year, however, the legitimacy of Broadbent’s theory was called into question. Employing Cherry’s dichotic listening task again, it was discovered that some words in the unattended message
are
processed for meaning. For example, if the name of an experimental subject is placed in the unattended message it is likely to be noticed.
Most people have had a direct experience of this phenomenon. If someone mentions your name at a party it tends to rise above the background noise, even if spoken quite softly. The background noise is the equivalent of an unattended message in a dichotic listening task. You are not conscious of monitoring the background noise, but to some extent, you must be – or you wouldn’t hear your name.
The finding that some words in the unattended message are processed for meaning suggested that the brain’s filter operates a late selection, rather than an early selection principle. All of the information arriving through the unattended channel receives a semantic analysis, to the extent that personal names can be identified, tagged as important, and subsequently admitted into awareness.