LECTURE 8 NOTES -MODELS OF ATTENTION
Under certain condns, we are very good at space perception—how come?
Perceive things globally opposed to details. Certain dimensions that globally analyze pictures instead of the details. Because we are
able to extract global info like openness, simple algorithms can be written in comp to extract taht from, then we have multiple
dimensions in which we can group pictures of any time.so that when we look at them they are semantically related.
Why is it that some times we have great difficulties, when?
-when same picture but not quite the same details. In terms of global features, the two pics are very similar.its impossible to
So when it comes to details, we have to use other strategies to tell the differences. They are local. We don’t represent them in first
glance but we are able to use attention and zoom in on those details and process them one at a time.and is very slow.
What explains the difference?
Last week lec fous was: process of relevant inputs, thoughts and actions.what happens when attention takes control of soemthing
This week lec: the cognitive mechanism that control what is going to enhanced or inhibited. What processes actually control
5 models of attention:
EARLY VS LATE SELECTION
Executive functions –are those that help us organize our behaviour.
Through motor area, we see RESPONSE
Defines processes of human beings. Includes everything except memory.
Essentially humanbeings are defined by sensorymotor integration.
Attn can play a role at any of these stages. Arrows indicate potential areas where attention may play a role.
Early selection –focus on certain modality sensory input and ignore others
Late selection- even semantic processes can remain unattended.
Most likely we have mechanisms that work on multiple levels so attn perform multilevel filtering
Perceptual load – depending on how much cog resources u have to devote to certain tasks, the more u are likely to ignore
-bold signal as a function of time rises while seeing motion.recorded from MT. Increase in bold signal only if u don’t do something
-People who had high load, the bold signal is showing barely any response for motion.infact, no motion has more of aresponse
~depending on how busy or not busy you are, you might or not filter something out.
Attention works like a spotlight or search light
Better perception for the process in spotlight
Attention s confined to a region in space and can move from one point to next.
Similar to eye movts
ZOOM LENS MODEL
You can zoom in to detail if u wish to i.e comparison of lines to determine which is higher or look at the wider scope.
PREDICTIONS USIGN INFO FROM SPOTLIGHT OR ZOOMLENS MODEL
Attention decreases on one side and increases on the other side without anything happening in between.
Not consistent with spotlight model.
2 problem:attention shouldnot split into more than one focus
Spotlight model predicts that we have one coherent region where attention is situated and confined. Can’t focu on soemthign else.
But this is possible.
Bunch of letters flashing at this red circles. 5 RSVP’s at the same time. Task: fixate at centre. Top left and top right. And told to ignore the rest( 3 diagonal dots)
There are better at detecting letters in both the top left and bottom right. ~we can pay attn to two areas and do poorly at the
But with spotlight model, the spotlight would overlap with the centre and would have shown improvement at the centre.
~its possible to split your attention and inconsistent with spotlight model.
Images; visual cortex for the left hemisphere and the right hemisphere. BLUE : decrease in bold signal. Red and yellow; increase in
LH shows the part of the visual field corresponding to the right bottom field
RH shows the part of the visual field corresponding to the left top field
~right becomes left and left becomes right
Multiple object tracking
- Multiple items on the screen. Some of the items will flash.blinking disappears. All the items start to move around.if u had
single spotlight of attention, to be able to track prior flashing items would mean that the whole screen would be the
spotlight and u would know where the distracters are as well.
- We can do multiple object tracking
Zoomlight and spotlight models are inadequate in explaining attention.
BIASED MODEL OF ATTENTION
Competition- u cannot process everything for a complex scene and have to focus on one thing at a time.
-all stimuli that come through compete for attention and processing
-limited resources in processing input and performing certain behaviour.i.e juggling, need many arms.
U need biases and need competition to be biased toward one stimuli.
Or without comptetition it would be confusing
These biases can be performed bottom up or top down. Illustrated inthe iamge
Salience: input that is brighter or have contrast.happens in vis. processing
The strongest person near the doorway would get through the door way at first.
-is flexible and not confined to certain region.
But who does the biasing?
PREMOTOR THEORY OF ATTENTION
You shift your attention because u have the occulomotor program ready to move your eyes.
Attention shifts happen before the eyes move.
Support for premotor theory of attention
Fixating on the left of the numbers. Mask is 88888
At soem point the 8’s changed into some number briefly
Undave to report the numbers.What number was at position at 2?
2 task; make eye movt
Colored ovals indicate the target to fixate
saccade to be made to one of the two colored ovals.
As u are about to make a saccade movt,there is some dealy.during this period, u are being probed with these letters.
Results: if asked about the letter at the saccade target, u are more accurate than anywhere else.
~ u can’t untangle eyemovts and attention.
Slide 8 shows the human occipital cortx
Outlines which areas in the brain that are associated with shifting attention.
IPS- intraparietal sulcus
PrCes – PRECENTRAL SULCUS
Teh bottom images: Left: when ppl shift their attn
Making eye movts
More or less the same. Evry similar regions activated. Lots of overlap. Must be the same process.
2 fronto-parietal networks that controls attention: frontal and parietal activation
Endogenous means volunteer shifts of attention.
i.e an arrow in the centre of screen. some thing representing spatial information for future stimulus.
Fixate dot on screen. Wrapped around the dot is some cue.i.e L . pay attention to box on left side.
After soem delay, faint target..delay. REP means report the side of the target.
P; means don’t have to shift your attention.keep fixated on the centre.but still have a letter to process.
The processes involve working memory areas but can subtract them and dont play a role.
Shift: Increased bold activation in medial and lateral areas in frontal cortx
and parietal.bilateral. left and right sides of hemispheres equally
contralateral wiring of brain: so L means activation in the right hemisphere etc.
Controlled cue; with P on screen: similar areas activated but more lateral part of frontal cortex.
So when u subtract
-other boxes become significant
Results: more medial activation.frontal eyefields. IPS, frontal and parietal cortex.
Shift: when L or R.
Skull activation changes
P; with in 350 ms, the rows look very similar.
These are probably to do with visual processing and working memory.