Words: 1563
Research carried out on attention has mainly been associated with the selective processing of incoming sensory information. It proposes, to some degree, our awareness of the world depends on what we choose to focus on and not simply the stimulation received by our senses. Attention is often linked to a filter that screens out most potential stimuli whilst allowing a select few to pass through into our conscious awareness, however, a great deal of debate has been devoted to where the filter is situated in the information processing chain (Martindale, 1991).
Psychologists have made extensive contributions to this subject matter in the past century. Notable examples include Donald Broadbent’s filter theory of attention (1958), which set the agenda for most subsequent work, and Anne Treisman’s (1964) modifications of this description, the Attenuation Model (Driver J, 2001). This paper will look at the theories of Broadbent, Deutsch and Deutsch and Treisman and will evaluate their attentional selection models and the theorists attempt to pinpoint the location of the attentional filter.
Broadbent’s model advocated the early selection theory, whereas Treisman disagreed on the location of the attentional filter and favoured a more flexible approach. A further model, which was appropriately named late selection, differs from the early and flexible selection model and was instigated by theorists such as Deutsch and Deutsch (1963) and Norman (1968). The three theories have differing opinions mainly over whether the filtering takes place early or late in the information processing chain (Gross, 2005).
Broadbent’s, Treisman’s and Deutsch and Deutsch Models of Attention are all bottleneck models, in that they predict we cannot intentionally attend to all of our sensory input at the same time. Each of these models endeavours to explain how the information that passes through the bottleneck is selected. When an individual is presented with two or more simultaneous messages and is instructed to process and respond to just one of them, this process is known as attentional selection.
The most popular way to measure this is through the process of shadowing, a dichotic listening technique which involves the participant repeating aloud everything that is heard in one ear. Cherry (1953) used this technique first when studying the “cocktail party phenomenon” in which individuals manage to select one or two voices to listen to from multiple conversations taking place at the same time and in the same environment (Gross, 2005). Cherry’s participants “shadowed,” or instantly repeated, prose heard through one ear, whilst ignoring further speech which was being presented to the other ear.
All participants were able to identify prose from the unattended ear as human speech, but no one was able to give a description of words or phrases that was exposed to the unattended ear, or the fact the voice had changed from English to German (Wood, & Cowan, 1995). In an attempt to explain Cherry’s findings Broadbent (1958) conceived the “Early selection filter model”. Broadbent’s model suggests that individuals have a restricted capacity to process information. Broadbent states that information arrives through sensory channels and is retained for a brief period in the sensory memory.
It is proposed that two stimuli offered at the same time obtain access to the sensory buffer (Eysenck, & Keane, 2010). The selective filter is used to block unwanted input and let through only those messages that warrant full cognitive analysis (Lachman, Lachman, & Butterfield, 1979). Information is selected on the basis of its physical characteristics for further processing by being allowed to pass through the filter.
The general consensus is that, as humans, we have only a restricted capability of processing information, therefore, this filter is designed to prevent the information-processing system from becoming overloaded. These key features are acknowledged with the need for a short term buffer store that preceded the selective filter. This buffer was a temporary memory store in which the unselected information could be held in parallel for short periods of time (Styles, 2006). Information can remain in the buffer for processing at a later stage, if this is not the case, it corrodes and is then lost (Martindale, 1991).
It is only when the “information passes through the filter into the limited capacity channel, which is a serial processor, that it is identified” (Styles, 2006, p19). The indication is that therefore, selection from the parallel input is made at early levels of processing, and is consequently an “early selection model” (Styles, 2006). A number of theorist’s have, however, introduced models that completely reject Broadbent’s Filter model. An example is that of Deutsch and Deutsch’s attention model (Martindale, 1991).
Deutsch and Deutsch (1963) offer a model in which all stimuli are fully analysed, with the most significant message determining the response (Eyesnck, 2010). This theory proposes a bottleneck, as the filter is placed closer to the response end of the processing system. This model is recognized as a late-selection model (Deutsch & Deutsch, 1963). The Deutsch and Deutsch model was later modified by Norman (1968) who alleges that sensory inputs are processed routinely and instinctively before we consciously know about them.
Norman suggests that no signal is filtered out but all are processed to the point of activating the stored responses in memory (Hampson, & Morris, 1996). Broadbent’s and Deutsch and Deutsch models disagree on the location of the attentional filter and they both are intransigent in their opposing views. This led some theorists to conclude that the location of the attentional filter may be more flexible, rather than fixed as Broadbent and Deutsch and Deutsch have suggested. Triesman (1969) devised a compromise model with the crucial difference being that the filter attenuates rather than eradicates the unattended material (Martindale, 1991).
Treisman was in agreement with Broadbent, in that, there was a bottleneck, but disagrees with the location. Treisman’s experiments used the speech shadowing method, were participants were required to shadow a verbal message in one ear whilst in the unattended ear participants were exposed to a string of words which were not entire sentences, but in places contained meaningful phrases. At various points in the process there was a switch with the attended message continuing on the unattended channel.
Treisman concluded that participants often changed to following the unattended message when the switch occurred. Triesman explains these results by accepting that there is some meaningful analyse of unattended information. Triesman therefore concludes that all information that enters the buffer store receives a degree of semantic analysis (Parkin, 2000). Triesman suggested that rival information is scrutinized for other things other than just its physical properties, including syllable patterns, after grammatical structure and meaning are processed (Hampson & Morris, 1996).
According to Treisman the non shadowed message is not filtered out early, but the selective filter attenuates it or weakens it, which would mean that “a message that isn’t selected on the basis of its physical properties would not be rejected completely, but have its volume turned down” (Gross, 2005, p219). Overall the anneunated filter process is flexible, depending on the information being attended to, but it was also thought that some information such as your own name might be permanently facilitated (Parkin, 2000).
Broadbent, Deutsch and Deutsch and Treisman all had their theories tested, Broadbent’s filter theory tells us that the non shadowed message does not pass through the filter, and this suggests that because the non shadowed message is filtered out early according to its physical characteristics, its meaning should not be subject to any higher level analysis. This does not however explain the “cocktail party phenomenon” when conversing with someone in one part of a room, are attention switches if we hear our name mentioned in another part of the room.
Moray (1959) demonstrated this, he concluded that when a participants name was presented to the non shadowed ear, attention switched to that ear about one third of the time (Moray, 1959). Regardless of the support for Deutsch and Deutsch and Normans theory, Wilding (1982) argued that there is not as much known about non- attended messages than the theories claim (Wilding, 1982) However, more is known than can be explained by either Broadbent or Triesman’s models (Gross, 2005). Broadbent’s, Deutsch & Deutsch and Treisman’s models have all been significant in the understanding of attention.
They all disagreed somewhat in the location of the attentional filter in the information processing chain. Early research suggested that the information processing chain was limited in its capacity to execute multiple tasks. Broadbent’s model was extremely influential in this and suggests that information is passed through a protective filter early on in the information processing chain, on the basis of physical characteristics and only this “information was passed on to be indentified” (Styles, 2006, p 40).
In research, processing information that is presented in the non shadowed ear led to the proposal that all “information was pre- attentively” (Styles, 2006, p 40) analysed for meaning, thus only the most significant signals were passed onto the response stage, this would suggest that the filter is placed closer to the response end of the processing system (Deutsch & Deutsch, 1963).
Treisman presented a compromise theory in which the selective filter attenuates the non shadowed message and overall the process is a more flexible one. Numerous researchers have queried whether any “single purpose, limited capacity central processor can account for all the complexities of selective attention” (Gross, 11, p, 221). Perhaps then the main difficulty with examining something like attention is that it’s extremely challenging to observe, as it is an internal process.