Models of Attention
Lecture 8 – Perception & Cognition
Short answer question:
Under certain conditions we are very good at scene perception. How come? Why is that it sometimes we have
great difficulty? When...and what explains the difference?
➢ We perceive things globally rather than locally – so not paying attention to every single detail
➢ Assuming that we're able to extract global information like openness, spatial frequencies, etc – we have
multiple dimensions that we can use to categorize pictures. So we can sort scenes into categories based
on small features?
➢ Sometimes it's hard to distinguish two pictures based on globalness – we have to use small details to
distinguish between them. We need to use attention to zoom into details, which is a slow process.
➢ Selective attention is a cognitive brain mechanism that enables one to process relevant inputs, thoughts,
or actions while ignoring others that are less important, irrelevant or distracting.
Last week: processing of inputs
This week: cognitive brain mechanisms that enable......control and determine what will be
enhanced or inhibited.
➢ Models ofAttention (5)
Early vs. Late selection
Sensory inputs –> Sensory InputAnalysis –> High Level Analysis – >Executive functions
• Attention may be involved in all of these levels?
Through motor areas we see some kind of a response
This is human processing in a simplified nutshell
This sensory-motor integration defines a human being
We can show behaviour based on immediate stimuli but also memory
Includes everything such as hugging, language
Attention plays a role in many of these stages.
• Watching tv and at the same time, there's a phone ringing, and you're deciding to ignore the
• Turn off auditory input?
• Early selection: focus on certain modalities more than others
• At very late stages, there's evidence that semantic processes can be performed
Multi-level filtering – many mechanisms of attention
• Depending on how many cognitive resources you have to devote to a certain task the more
you might ignore everything else
• Ex: You might not perceive motion, even thought there is motion, just because you're also
doing other tasks? So the area in your brain responsible for detecting motion does not show
Attention works like a spotlight
You have a stage and people on it. You have one spotlight directed on one of these people This person should be more salient.
Spotlight (or search light model):Attention is confined to a coherent region of space and can
move from one point to the next.
Zoom lens model
Attention expands from fixation, grows to fill whole region, shrinks to include just cued location
Zoom into details or have a wider scope to take the whole scene in.
• Problem with Spotlight Model and Zoom Lens Model
• Attention should behave a certain way according to these models.
• 1 - These models presume that attention shifts
• If attention shifts from one person to another, then what happens is your spotlight
will shift to the other side passing an empty space (spotlight prediction)
• Zooms lens predicts something similar.
• So both models predict that you will pay attention to the empty space inbetween.
• What they found is that this isn't true. You don't shift your attention that way. It
ndjumps quite quickly so you don't focus on the empty space inbetween.
• 2 – Spotlight Model says attention should not split into more than 1 focus.
• But really you can split your attention spotlight.
• You can actually attend to two objects on the side while ignoring the center. But
according to spotlight, attention should be devoted to center as well.
• So we can split our spotlight of attention!
• We don't just focus our attention like a spotlight on one huge area in space. We can
focus on some of the objects in this area while ignoring others/
• Object tracking
Biased competition model
Contemporary Model (1995)
Desimone & Duncan
Competition: stimuli in the visual field that compete for limited processing capacity & control of
• You just can't process all the rich details of 2 different scenes at the same time (Change
• You will favor some stimuli over others
• Biases can be bottom-up or top-down
• Competition biased towards certain stimuli depending on
• bottom -up mechanisms: salience
• You process whatever stands out first
• top-down biases
• Can be based on anything (spatial cue, etc)
• So who's doing the biasing?
• Homonkulus model doesn't offer an explanation for this.
This model is backed up by evidence
Premotor theory of attention
1) There's a stict link btwn orienting of attention (cover attention) and programming explicit
occular movememnts (overt attention)
2)Attention is orientedto a given point when the oculomotor program for moving the eyes to this point is ready to be executed.
• Shift your attention because u have a motor program that's ready to move your eyes
3) Covert orienting of attention (w/o eye movememnts) is achieved by inhibiting the execution
of eye movement itself.
• You have the