Introduction

Attention is a complex topic concerning how the mind deals with sensory data. The very thing that makes attention so difficult is its use as the primary way of analysing consciousness. Allport (1980) described it as 'the experimental psychologist's code name for consciousness'. However, most interpretations define attention either in terms of the limit to the amount of processing that can be carried out (divided attention) or in terms of the mechanisms that reject or register information (selective attention). It is this latter selective approach that will be considered when dealing with the cocktail party effect, as selective attention concerns the focussing of awareness on a limited range of stimuli. Studies of attention began in World War 2, when people asked to observe radar for hours on end were found to have missed and estimated 30% of all targets.

Attention, in many ways, is similar to perception, especially when dealing with selective attention. This view is summed up in a quote of Greene and Hicks (1984), "We can only perceive things we are attending to; we can only attend to things we perceive."

The cocktail party effect, when a person's own name is heard and recognised in a channel, or conversation, that was not being paid attention to, is a perplexing phenomenon that serves as a criticism for many attentional models, thus making them void. An example of such a theory is Broadbent's filter model (1958). He proposed that attention is the result of a limited capacity information processing system, and that most channels are not paid attention to in order to cope with the sheer amount of sensory stimuli occurring at any one time. From the results of many dichotic listening tests, such as Cherry's 1953 study which found that some physical characteristics of an unattended message could be identified, such as the sex of the voice, Broadbent concluded that the filtering occurs due to physical characteristics. Many aspects were identified as being recognisable or not in an unattended message; those that were included volume, brightness, intensity and novelty, whereas those that could not be identified were the content and language.

Following, in 1964, Treisman developed the attenuator theory. This maintained that channels attended were selected due to physical characteristics, however instead of completely ignoring unattended channels, they were simply diminished and analysed semantically along with attended channels. While this theory does account for the cocktail party effect, it does have many limitations. These include a lack of explanation of the attenuation process, and also a failure to recognise that analysing every channel semantically is extremely inefficient, a criticism that also makes the pertinence model by Deutsch and Deutsch (1963) seem almost nonsensical. This proposes that all channels are semantically analysed and then passed to an attenuator. Channels attended to are those that are pertinent.

Back to the top

Conclusion

In conclusion, these theories fall into two categories, the first, i.e. Broadbent's theory, explain attention accurately, and fail only on explaining the cocktail party effect. The latter two models succeed in explaining this happening, yet in doing that become overly complex. Indeed the occurrence of the cocktail party effect provides a paradox in itself; channels need to be switched in order to detect the meaning , while the cocktail party effect sees channel switching caused by deciphering the meaning.

My proposal is that the cocktail party effect does not occur from semantic analysis but from physical. The name is a word learnt from a very early age, and is so strongly associated with the self and over such a long period of time, that it becomes recognised automatically, in a physical way. Merely the sound and form of the word is recognised as the name.

Back to the top

Evidence

Investigations (by Cherry etc.) do show that physical aspects are easily detected in a non attended message. However it is also necessary to discuss the subject of automaticity.

Automatic processing includes performing a task, such as hearing and recognising a name, automatically and without conscious attention (where as controlled processing requires conscious attention.) Studies do show that repeatedly practising a task requiring controlled processing (such as first learning the name) eventually causes it to become automatic, and indeed this is evident in everyday life when learning a new skill, e.g. driving a car. Schneider and Shiffrin's 1977 theory of automatic processing stated just this, yet one criticism was that it didn't explain how this occurred. In response to this it was suggested that the processing and retrieval of the appropriate response became quicker and quicker until it occurred almost instantaneously. Furthermore, three different levels of automaticity were proposed by Norman and Shallice (1980), along with detailed explanations of how these occur. These were; 1. fully automatic processing controlled by schemata with hardly any conscious awareness, 2. partially automatic processing involving contention scheduling as a way of resolving conflict between competing schemata, which uses some conscious awareness but virtually no conscious control, and finally 3. deliberate control governed by a supervisory attentional system using full consciousness and flexibility.

Automatic processing occurs quickly, unavoidably, and does not reduce the capacity for performing other tasks simultaneously. However, when observing the cocktail party situation, it can be seen that, if the recognition of the name is taken as an automatically processed reaction, then this task intruding into conscious attention contradicts the claim that it does not involve consciousness at all. It is when the stroop effect is considered that this proposition makes sense. This involves an automatic task disrupting the performance of a controlled task, where participants are shown colour names written in a conflicting colour of ink, and asked to either read the word or state the colour of ink used. When attempting to state the colour ink, the participant either takes longer to answer, hesitates, or begins to say the printed word rather than the colour of the ink. This suggests that, while the participant is automatically reading the word, this is intruding and hindering the performance of the task in hand.

Thus it can be seen that the cocktail party effect occurring from an automatic response to a physical recognition is one that does have it's theoretical evidence. It is now apparent that this must be tested experimentally.

Eysenck (1984) identified two main methods of studying attention. The first is to present two or more simultaneous messages to a participant and to instruct them to respond to only one using the 'shadowing' method of repeating a message as it is heard. This is sometimes known as a dichotic listening test, when two different messages are presented to different ears. This technique concerns selective attention, and will be the procedure used for this investigation. Secondly, participants are asked to respond to both or all messages, requiring divided attention, which this experiment does not concern.

Back to the top


Joanne Pilkington 2000