Broadbent’s filter model

Broadbent’s filer model is a theory of attention that states that we have limited resources of
attention and therefor our attention system works like a filter. Information enters our attention
system enters in a successive manner. This limited capacity of our system has a bottleneck proto
type. The bottleneck prevents the fluid from entering the bottle's body all at once; instead, it allows
the fluid to enter in increments based on the flow rate until the bottle's body is completely filled. To
attend deliberately to sensory stimuli in a way that prevents sensory overload, we need to practise
selective attention. This filtering of the information takes place at a very early stage of information
processing. There’s no semanticity and it only takes place on the basis of physical characteristics.
According to Broadbent, our information filtering starts out very early in the perceptual process.
Later operations used physical features such as colours, loudness, or directions of the stimulants
processed earlier to accept or reject a stimulus.

When constructing his model, Broadbent placed a strong emphasis on the division of incoming
inputs into attended and unattended channels. Attention directs the choice of channels. Reflexive
attention is used when a sensory event catches one's attention, as opposed to voluntary attention,
which is used when one is trying to attend to a stimulus based on their present goals. After being
chosen to pass through the filter, the information can then be stored in short-term memory and
modified before being stored in long-term memory.

For example, to prove the idea of this attentional process, Broadbent applied the dichotic listening
test. This task has been used widely to examine a variety of psychological phenomena, including
testing for attended and unattended information provided to a participant as well as response times
to specific auditory information. It is popular because it is a non-intrusive way to assess brain
dominance. In a standard dichotic listening paradigm, the participant wears headphones and their
attention is split between two separate auditory stimuli that are presented simultaneously to each
ear. The subject is told to ignore (unattended channel) the information presented from the other ear
piece while paying attention to the information coming from the other ear piece. The participants
are put to the test to see if they can recollect any of the information that was delivered in the
unattended channel after the listening period.

Early studies that employed dichotic listening tasks produced empirical proof that participants had
better recollection while the channel was being attended and less recall when it was being ignored.
The participants are put to the test to see if they can recollect any of the information that was
delivered in the unattended channel after the listening period. In his split-scan tests, Broadbent
employed this paradigm, giving subjects a separate letter in each ear at the same time and asking
them to repeat it in any sequence. As a result, the letters presented to one ear were reported first,
followed by the ones provided to the other ear. Participants filtered the information according on
ear, which supports Broadbent's filter theory and an early selection model.

Attenuation theory

The cocktail party effect happens when you are listening to and paying attention to one message but
also hear portions of another essential message that is not being paid attention to, such as your
name or other important phrases like "fire!" As a result of the cocktail party effect, Anne Treisman
modified Broadbent's early selection model of attention and coined the term "attenuation theory of
attention." The filter in Broadbent's theory is replaced by an attenuator in the first step of the
attenuation theory's two-stage selection process, and a dictionary unit is used in the second stage.

Treisman showed in her research that participants could still recognise the contents of an
unattended message, demonstrating that they could process the meaning of both attended and
unattended messages. The selective filter in Treisman's attenuation model selects between two
messages based on their physical properties, such as position, intensity, and pitch. Treisman's
concept includes a "dictionary" that enables messages to be chosen based on their content. In order
to activate awareness of a stimulus, some information, like our name in the cocktail party example,
requires a relatively low threshold. Therefore, the attenuation model suggests that the perceived
loudness of an unattended message will diminish. Unless it has a very low threshold to begin with
(your name), or there is a general transitory decline for all messages, this message will typically not
be loud enough to hit its threshold.

Figure 1: triesman's attenuation model of attention

For example,

As part of the psycholinguistic experimentation method known as "speech shadowing," participants

During the shadowing tests, Treisman would give each ear a different stream of banal stimuli. The
stimulus would then switch sides at some point during shadowing, presenting the previously
shadowed message to the unattended ear. Before realising their error, participants would frequently
"follow" the message to the unattended ear, especially if the stimuli had a high degree of continuity.
Contrary to Broadbent's filter model, which predicts that participants will be fully unaware of the
change in the unattended channel, this "following of the message" demonstrates how the
unattended ear is still extracting some information from the unattended channel.