This is the 2007 version. Click here for the 2017 chapter 07 table of contents.

Selective Attention

As a rule, attention cannot be focused on more than one thing at a time. It has a limited capacity. This concept was discussed in connection with working memory in Chapter 6. Working memory is, for all practical purposes, a synonym for focused consciousness or attention. George Miller discovered in the 1950s that primary memory could hold about seven (plus or minus two) items at the same time. Working memory therefore acts like a limited capacity store. How does the cognitive system determine which material is let into this limited storage area for intensive processing? That question constitutes the problem of selective attention.

What is the "cocktail party phenomenon"?

In the 1950s, Donald Broadbent came up with a filter theory of selective attention. He proposed that the human cognitive system had some mechanism (a filter) that could remove unwanted stimuli from attention. This filter theory seemed to receive support from classic research into the cocktail party phenomenon by E. Colin Cherry. Cherry pointed out that, at a party, we can tune in on one conversation while tuning out others.

Further studies showed that the brain uses directional information to help separate conversations. If voices come from different directions, we easily tune one "in" and the other "out." (The direction of sound is determined by a variety of factors, such as different arrival times at the two ears.)

What is evidence that part of the brain pays attention to non-attended messages?

Cherry also found evidence that filtering was not complete. If one's name is mentioned in a conversation to which one was not paying attention, on the other side of the room, one might nevertheless notice it. Evidently some part of the brain monitors information to which we are not paying attention.

Similar findings came out of experiments on dichotic listening by Anne Treisman (1960). Treisman put headphones on her subjects, with different messages being played to each ear. She instructed her subjects to pay attention only to the message in one ear, not the other. Subjects could easily do this. Treisman had the subjects shadow the message, which meant to repeat the message out loud, word for word. That proved they were listening to it. Subjects had no trouble shadowing a message that was played to one ear while a different message (in the same voice) was played to the other ear.

What did Treisman discover, in the dichotic listening experiments?

Triesman found that if she suddenly switched the messages to the opposite ears, in the middle of a sentence, the subjects "crossed over" to other ear for a few seconds to complete the sentence. For example, if the voice going into one ear said, "To be or not to be..." and a moment later the same voice, in the other ear, said "...that is the question," subjects would shadow the whole sentence, which violated their instructions because they were supposed to pay attention only to one ear. Treisman thereby confirmed what Cherry suspected: subjects do monitor the information in the unattended ear, even if they screen most of it out.

How did Mackay document an effect from messages in the unattended ear?

After Treisman published this finding, other researchers confirmed that information in the nonattended ear could influence the thought process in various ways. For example, Mackay found that information in the nonattended ear could disambiguate a word presented to the attended ear. If the word in the right ear (which the person was shadowing) was "bat" and the voice in the unattended ear was talking about caves, "bat" was interpreted as referring to a flying mammal, not a wooden bat.

How did Broadbent modify his theory in the 1960s?

Findings like this led to the demise of the filter theory. If words going into the nonattended ear were being analyzed for meaning, then clearly there was very sophisticated processing (capable of analyzing linguistic meaning) going on outside attention. Broadbent recognized this when he modified his theory in the 1960s. Now he hypothesized a parallel process that came before attention. Its purpose was to direct attention toward something that might require more conscious processing. When we discussed driving on autopilot, at the beginning of Chapter 3, we mentioned something similar. A driver will "snap out" of the autopilot state if anything unusual happens. Broadbent was proposing a similar process.

What did Shiffrin and Schneider conclude? How is the "pop-out" phenomenon relevant to their conclusion?

Shiffrin and Schneider (1977) came to a similar conclusion after analyzing visual search experiments. When a person scans a long list of items, something different or unusual will draw itself to attention automatically. Shiffrin and Schneider referred to this as the "pop-out" phenomenon. For example, the letter "O" in a page full of X's will just pop out. One can hardly avoid noticing it. Shiffrin and Schneider summarized this and much other evidence pointing toward two types of processing in human cognition, controlled processing and automatic processing. Automatic processing calls attention to patterns which might be important, among other functions. Controlled processing is used to evaluate the situation in more detail.

This might remind you of the two modes of processing noted by Neisser (who pointed out in 1963 that this idea was repeatedly discovered by different researchers). One type was automatic and parallel, the other type was controlled and conscious. Shiffrin and Schneider discovered the distinction again.

Write to Dr. Dewey at

Don't see what you need? Psych Web has over 1,000 pages, so it may be elsewhere on the site. Do a site-specific Google search using the box below.

Custom Search