Dstinction Between Early-Selection and Late Selection in Models of Attention

Attention is the ability to take a hold of the mind and select one from many simultaneous objects or thoughts, discarding some things in order to deal with others more effectively (James,1890, as cited in Norman, 1976) . In every day life there are so many things that can be attended to and so certain aspects have to be selected over others. Some tasks that are carried out require very little attention and others require less the more time that they are practiced.

All sensory information that is received requires some cognitive processing however with attention, cognitive processing needs to be continuous in order to keep focus and concentration. Attention is something that could be said to be limited, as cognitive processing cannot occur for all stimuli at the same time. It also is selective as the mental energy required for attention can be expanded as the person sees fit. Attention is an issue that has been debated upon for many years.

Early-selection models have looked at selection as a limited process that requires selective attention in order to proceed.

The selective attention occurs after a basic analysis of the physical features that are used to tell the difference between selected and non-selected stimuli. As a result of this, the unattended stimulus is not completely perceived. Late selection models however discard this idea and assume that perception is unlimited and that attention can be performed in parallel without a need for selection. Lavie and Tsal, 1994) Both the late selection models and early selection models differ in their concepts of how people attend to things and it was only until a compromise model was put forward by Lavie and colleagues that both models were combined to form the perceptual load framework.

Get quality help now

Bella Hamilton

Verified

Proficient in: Metaphysics

5 (234)

“ Very organized ,I enjoyed and Loved every bit of our professional interaction ”

+84 relevant experts are online

Hire writer

The idea of how we attend to more than one thing at a time has always been an issue. Colin Cherry (1953, as cited in Eysenck and Keane, 2001) looked at this idea when at a party he was fascinated with how people were able to follow just one conversation when there are many other conversations taking place around.

He called this idea the ‘cocktail party effect’, and put it down to physical differences, like gender, speaker location and voice intensity. Cherry carried out a number experiments on this idea and did this by presenting subjects with two different messages at the same time, one in each ear. Cherry found that when the messages were of the same voice then the listener was unable to distinguish between either of the two different messages. Cherry then carried out studies in which he asked the listeners to ‘shadow’ one of the two messages heard.

He found that the listeners could extract very little information from the non-shadowed message, and barely noticed when the non-shadowed message changed language or was even read backwards. Moray (1959, as cited in Kahneman, 1973) tried to look at how much information a subject could remember in the rejected auditory channel and so repeated an english word 35 times in one of the ears. He found that even though the subjects had heard the word many times, many of them still could not retain what it was.

Cherry concluded from his work that unattended information receives hardly any processing, which follows the early selection model ideas that attention is not processed in parallel. Broadbent (1958, as cited in Eysenck and Keane, 2001) developed Cherry’s ideas further by conducting an experiment on participants in which he presented them with two sets of numbers in each ear. For example the number 354 was presented in one ear at the same time as the number 692 was presented in the other ear. Broadbent found that participants recalled the numbers ear by ear rather than as a whole number, for example 354692 instead of 365942.

Broadbent explained his findings by saying that both the inputs are presented in parallel fashion until they reach a buffer where one input is allowed to pass through and the other remains for later processing. This he explained is to prevent an overload. From his discovery Broadbent went on to construct a model from which he felt best described how attention is processed. This model was called the filter model. The filter model put forward the idea that the ability to analyse and identify information is limited.

He argued that when information reaches the brain it filters out desired messages and rejects everything else, therefore blocking undesired inputs and preventing an overload on the perceptual system. (Dittrich, 2004) When information is presented it is analysed based upon its physical characteristics. The information is then simultaneously stored in separate channels in the sensory register and the filter switches between channels passing the information stored there, one channel at a time, to the detection device. If the message coming in doesn’t have these characteristics then it is filtered out.

Broadbent’s ideas were the beginning of many years of further research. This begun with a referral back to Cherry’s questioning of how people can switch their attention from one input to another when they are unaware of the content within the unselected input. Gray and Wedderburn (1960 as cited in Kahneman, 1973) looked at this and ignored Broadbent’s idea that attention was the basis of an analysis of physical characteristics. They believed that psychological aspects played an important part and so carried out experiments to try and prove their theory.

The experiment involved presenting subjects with words broken down into syllables. Each syllable was presented alternately to each ear. At the same time another word was broken down the same way and was presented to the complimentary ear. Gray and Wedderburn found that words could be constructed and were not repeated as just a mixture of different syllables. This suggested that meaning of information can be constructed from both ears and it is not just physical characteristics like Broadbent had previously said.

Continuing from the work of Cherry and Broadbent, Treisman (1960) discovered that in experiments where a participant had to shadow a word, they sometimes repeated a word that had been presented to the unattended ear. This was known as a “breakthrough”(Dittrich, 2004). Treisman carried out a series of experiments and found that on the cases where this did occur the majority of the time the word was related to the context of the words being presented to the attended ear. Findings from this led to Treisman proposing a new filter theory in 1968.

The filter rejected the idea of Broadbent’s all or none filter idea and suggested that attention is processed in a more flexible manner. Treisman noticed that a difference in voice and language played an important part in experiments involving participants to select a response to one of two messages ( Norman, 1976). It seems that differences occur depending on what language the second message is spoken in. For example if the language is of similar phonetic structure to that of English then it is less distracting than if the language was something like Czech.

Also if the subject has previous knowledge of a language then this will be more distracting than hearing a language that they have never heard before. Overall Treisman concluded that it is unlikely that both messages are fully analysed and so selection of one message takes place over the other This selection takes place starting with an analysis of the messages physical features, then goes on to look at words and meaning. Triesman said that selection of messages that have the same voice, intensity and localization takes place during the analysis rather than after it.

It is at this stage that the information-handling capacity becomes limited and so to prevent overloading the irrelevant message is either discarded or both messages are kept and alternation takes place between each one. The decision to which message is kept or whether both are alternated is based on physical characteristics and grammatical features. This is why people are sometimes able to anticipate what words are about to occur when speaking to someone about a particular subject.

It is also why that information from competing messages is sometimes picked out when the information being heard is similar to that in which is being primarily attended. Deutsch and Deutsch (1963, as cited in Eysenck and Keane, 2001, Norman, 1976) come up with a theory, which is known more as a late selection model. According to Deutsch and Deutsch all information is processed in full when it enters the consciousness and that it will reach the same perceptual and discriminatory mechanisms whether attention is paid to it or not.

They believed that sensory stimulus does not affect what we attend to but each central structure has different levels of importance. For example, our own name makes us attend more or distracts us more easily from a primary source than any random word. Deutsch and Deutsch disregarded Triesman’s and Broadbent’s early filter theories and put forward the idea that even if a signal is on a channel in which another signal is already being attended it still can be detected.

They also suggested that all inputs are recognized in parallel, and that the role of attention is to select which items get placed in short term memory. There is a lack of evidence supporting Deutsch and Deutsch’s theory and competing theorists have provided evidence against the idea that all information is processed in full. For example Treisman and Geffen (1967) asked participants to shadow one of two auditory messages and tap when they heard a target word in either of the messages.

Results found that detection of the target word was much higher in the shadowed message than the non-shadowed message disregarding Deutsch and Deutsch’s claim that there should be no difference in the detection rate. Lavie et al (1994) took the ideas of both the early and late selection models and came up with a compromise of the two; this model was called the perceptual load framework. Lavie et al proposed that although physical differences play an important part in selective attention, irrelevant information is only discarded when the perceptual load of relevant information is too high to make use of the available resources.

Therefore Lavie et al argues that tasks have different levels of attentional capacity and that upper limits are imposed at different levels dependent on a persons temporal state of alertness, availability of resources and the subject-task structural constraints. Navon and Gopher (1979 as cited in Lavie and Tsal, 1994) came up with a rule that states the amount of resources that can be applied to a task. They said that resources will be supplied to meet the internal demand and that these resources will either equal the demand or reach the limit of the available resources.

Lavie et al’s approach mixes both the early and late selection processes by looking at the distinction between high and low perceptual loads. A perceptual load is the number of units in a display and the type of processing that is needed for each unit. It seems that the results that support early selection models have been gained when the conditions have been of high perceptual load whereas results supporting late selection have been gained under conditions of low perceptual load. As well as all of this Lavie et al still takes into consideration the differences between the physical characteristics of attended and unattended information.

An experiment carried out by Stroop (1935 as cited in Lavie and Tsal, 1993) shows that there is a poor physical distinction between relevant and irrelevant stimuli when they both occupy the same location. This was shown in an experiment in which subjects had to identify the colour of a word. Results found that subjects processed the meaning of the word and that reaction time increased in relation to the baseline of response time to recognising a colour patch or the colour of a non-colour word when the word was a colour name and printed in a different colour.

The stroop effect therefore was said to show that one stimuli affects another more one way, than if the two stimuli were to be reversed. For example naming a colour doesn’t usually affect word reading, but word reading can affect colour naming. This indicates that the stoop affect is an example of the difference in parallel processing between relevant and irrelevant stimuli, which shows why there is a difference in response times between some stimuli compared to others.

In all this approach looks at perceptual load being of a low level and generally supports late selection models as it takes upon the idea that attention is a process that can be carried out in parallel. Looking at evidence supporting early selection models however takes into consideration high perceptual loads. Pashler (1984, as cited in Lavie and Tsal, 1994) devised a variation of the bar-probe technique (Averbach & Coriell, 1961).

This involved using eight different items shown on a display with three stimulus onset asynchronies being shown between each stimulus and the probe. Subjects had to match each of the four shapes with the location, which resulted in the subjects creating eight different combinations. Pashler found that physical differences increased the time it took the participant to identify each shape to the location, even when the probe came 300 msec before the display. This therefore questions the late selection models as it shows that selection happens at the perceptual stage.

In conclusion it seems that there is evidence supporting both models of early and late selection and that attention is something that varies dependent on a combination of both ideas. Although Broadbent (1958) and his filter theory initially put forward the idea that attention is a selective process, it does seem from later research that in some cases people are able to attend to more than one thing at a time. Treisman (1960) developed Broadbent’s idea and extended his filter theory by suggesting that physical differences play an important role in determining which message to keep or discarded.

Treisman believed that selection takes place before analysis rather than after it and that both messages can be kept but alternation has to take place to prevent an overload. A late selection model proposed by Deutsch and Deutsch (1963) disagreed with that of early selection models in that they believed that all information is processed in full when it enters the consciousness. Deutsch and Deutsch said that even when multiple messages are on the same channel they can still be processed in parallel with the most important being kept and the other being stored in the short-term memory.

Comparisons of the early and late selection models finally resulted in both being examined in an equal manner, eventually leading to the perceptual framework model being proposed by Lavie and colleagues (1994). This model supported both selection models by linking it to whether a perceptual load was high or whether it was low. Overall it seems that attention is wholly down to an analysis of physical properties, which is then given further analysis to determine what things are paid attention to and what things are not.