There are two dominant models of attention that have been proposed: early selection and late selection. The early selection model posits that attention operates as a filtering mechanism, with only information that is relevant to the task at hand being allowed through to conscious awareness.
The late selection model, on the other hand, argues that all information is initially processed in an undifferentiated manner, but that only information that receives further processing due to its relevance to the task will be brought into conscious awareness.
Both models have received support from empirical evidence, but the debate between them is far from resolved. In this essay, I will compare and contrast the two models, looking at both the evidence in support of each one as well as the criticisms levelled against them. I will argue that, while the early selection model is more parsimonious, the late selection model is more in line with our current understanding of attentional processing.
William James (1890) defined attention as “possession of the mind, in clear and vivid form, of one out of what appear to be several conceivable objects or trains of thought. Its essence is focalization and concentration of consciousness.
Attention has been the focus of much research within psychology and cognitive science, with numerous theories and models being proposed to explain how and why we attend to certain stimuli in our environment. In this essay, I will compare and contrast two of the most influential Attentional Selection Models – the Early Selection Model and Late Selection Model.
The early selection model of attention was first proposed by Broadbent (1958, cited in Eysenck & Keane, 2000), who suggested that Attention acts as a filter, screening out unimportant information and only allowing relevant information through to be processed further. According to Broadbent, this filtering process occurs very early on in the processing of information, before the content of the message has been fully decoded.
Broadbent’s model was later expanded upon by Deutsch and Deutsch (1963, cited in Eysenck & Keane, 2000), who proposed the notion of ‘selective Attention’ – the idea that we are able to focus our Attention on one particular message whilst still being aware of other messages in our environment.
The late selection model of Attention was first proposed by Neisser (1967, cited in Eysenck & Keane, 2000). Neisser’s model suggests that all information entering our awareness is fully processed, regardless of whether it is relevant or not. However, he also proposed that certain types of information may be given preferential treatment during processing – for example, information which is self-relevant or which has emotional significance.
Both the early and late selection models of Attention have been supported by empirical evidence. For example, Broadbent’s (1958) filter theory was supported by the findings of his ‘shadowing task’, in which participants were asked to shadow (repeat aloud) one message whilst ignoring a second, competing message.
Broadbent found that participants were only able to accurately shadow the attended message when it was presented early on in the sequence – when they had time to process it before the second message began. This finding suggested that information which is presented later on is not given any preferential treatment during processing and is instead filtered out.
The findings of Deutsch and Deutsch’s (1963) ‘selective Attention task’ also supported the early selection model of Attention. In this task, participants were asked to listen to a tape-recording of two messages, one presented in each ear. They were then asked to count the number of times a particular word was said in one message (the ‘counting condition’), whilst ignoring the other message.
Deutsch and Deutsch found that participants made more counting errors when the words to be counted were presented in the same ear as the ignored message, suggesting that they had not been filtered out as intended.
Neisser’s (1967) late selection model of Attention was supported by his finding that participants were better able to remember self-relevant information than information which was not self-relevant. This finding suggested that self-relevant information is given preferential treatment during processing, even when it is presented alongside other, competing information.
Overall, both the early and late selection models of Attention have been supported by empirical evidence. However, it should be noted that the two models are not mutually exclusive – it is likely that Attentional Selection occurs on both an early and late basis, depending on the nature of the stimulus and the individual’s goals and motivation.
This definition emphasizes how selective attention is regarded. It appears self-evident that we can’t pay attention to all stimuli at the same time, thus some method of selection must be used to determine what information we focus on and process further, and what is disregarded.
The terms early and late selection models of attention refer to when this process of selection occurs in relation to the rest of information processing.
The early selection model, first proposed by Broadbent (1958), suggests that selection takes place very early on in the processing of information, before any detailed analysis has taken place. In other words, we filter out unimportant information before it has a chance to reach our consciousness. This model is supported by research which shows that we are good at ignoring irrelevant information when carrying out tasks which require our attention (e.g. Cherry, 1953).
Since the 1950s, much study has been done on selective attention, both auditory and visual. Several distinct theories and models of selective attention have been developed over the years. The question of whether selection happens early or late in processing has long divided researchers.
The early selection models propose that selection takes place at the level of perceptual input, before any further processing has occurred. In other words, attentional resources are deployed to certain stimuli and ignored others before any meaningful or semantic processing of the stimuli has taken place. The most famous model of early selection is the Attentional Filter model proposed by Broadbent (1958).
According to this model, information from all incoming sensory modalities is first sampled indiscriminately before being filtered according to certain characteristics (e.g. loudness in the case of auditory input) in order to focus on a particular stimulus. This filtering process is assumed to be relatively permanent and unselective – i.e. the same filter will be applied regardless of whether the task at hand is relevant to the filtered characteristics or not.
In contrast, late selection models propose that attentional resources are not deployed until after a certain amount of perceptual and/or semantic processing has occurred. Selection therefore takes place at a higher level of processing, when the stimulus has already been interpreted in some way.
The most famous model of late selection is the cocktail party phenomenon proposed by Cherry (1953). According to this model, we are able to focus on one particular conversation (the ‘cocktail party’) even though there are many other conversations going on around us (the ‘background noise’). This is because we can semantically interpret the conversations and select only those that are relevant to us.