Paul- Attention Part 1 Flashcards
(22 cards)
Background- William James
Expereince is about attending to things. Attention is focusing on one thing whilst suppressing others
Early selection models- broadbents filter model
Info not attended to is filtered out before processing. Structural but not capacity limitations. Sensory store for both attended and not, then filter, then only attended by perceptual processing, then stm and response . All or nothing, no multitasking as it’s one after the other
Cherry- evidence for broad bent filter model
Dichotomy listening task. For unattended could only tell physical features e.g. gender, reversed, language but not content. Only input from one channel processed at a time, when told to switch there was a delay so lost info
Moray, treisman disputing broadbent filter model
Moray: gave shock associated to word, when heard in unattended, had galvanic skin response and words w similar meanings so processed semantically (unattended processed fully do need attenuation). Treisman : started sentence in attended then moved, ps followed the sentence. Created attenuation model unattended turned down, weaker signal but some louder than others like primed (name). Attenuation happens priori to processing, structural, sensory first
Late selection models: Deutsch and deutsch
All info processed to the highest level, all physical processed, named and categorised. Meanings extracted in parallel. Selection occurs at level of awareness. In flow chart: everything into sensory, perception, then filter after processing for stm or not- structural
Evdeince for late selection- lackner and garrett
Ps use material from unattended to resolve ambiguous sentences. In stroop task, word and colour are processed and compete. One has automatic (reading) and one controlled (naming)- all info encoded then filtering after to select response
Weaknesses of evidence
Unattended info may be remembered as filter leaky soall info processed. Also attention may slip and attend to stuff told not to- this links to early selection. Evidence from 1 comes from experiments that didn’t ensure info really attended
Capacity models of attention
No structure limits, cog system has limited processing capacity so attention is a resource. Depends on task difficulty, expertise. Allocation depends on enduring disposition, momentary need for right now, evaluations of demands on capacity
Kahneman - capacity model
Too much arousal means reduced focused, notice more irrelevant details. Too little means no motivation. Increased levels can counteract capacity limits are makes you more focused. Did dual task, first separate the. Together. Drop in performance called dual task decrement.
Dual task evidence
Ppl can learn to do 2 things at once- spelke- reading prose while writing, after 6 weeks went from 0 to competence
Treisman and davies 73: ps monitor two screens, two auditory inputs and one of each. All showed decrement, 1 and 2 had worse performance compared to 3 so resource may be modality spec
Perceptual load theory-Lavie
System is limited like early selections but processes everything until runs out of capacity. Low demand-more or what is seen is processed-increased chance of attending to irrelevant information . High demand-less is processed-non relevant filtered out. Can have early or late selection depending on demands. Evidence: low load: strong compatibility effect e.g. when distractor incompatible, do have interference. But no effect in high, no interference as distractor not being processed. Low means late filtering (more processed),
Inattentional blindness cartwright finch and lavie 2006
Not noticing task irrelevant elements
Load reduces compatibility effect, increases blindness- ps either asked colour/easy or size/harder of line, also had irrelevant- square in corner. Simple saw square. Hard didnt
Wm
Ps given set of digits to remember, low is in order, high is diff each time. Then asked to sort faces w cong or incong names and probe digit
High wm load increases impact of incong pictures- bigger diff in rt between cong and in compared to low
Overloading passive reduces interference, overloading active increases it
Visual search initial paradigms
Treisman: ps have to find a target amongst distractions, vary the number of distractions (size of array), presence or absence of target (+ or - trials) and either feature search (searching for one thing) or conjunction search )e.g. orientation and colour) for positive feature search, rt doesn’t increase w no. Of distractors but does for -. For conjoined, some of array does increase for + and even more for -
Slope equation for searching activities
Y=mx+c where c is time for processing time, m/slope is the ms each item adds to the search and x is no. Items. Rt=ms per item times number of items+ duration of non search processes.. slopes for conjunctive - 67, conjunctive + 28, single + 3 single - 25. E.g. as you increase array increase search time by x. E.g RT increase for every item added. Slope steeper for - as have serial search to check for target but not for + as jumps out/parallel
Feature integration theory - treisman and Schmidt
Two stages: initial stage feature detection: absorbs features quickly, parallel, efficient, pre attention, single features pop out so all you need for feature search. Stage 2 is feature integration where you use attention and serial processing, used for conjoined search, bring elements together, slow.
What slop data should look like vs what it actually looks like
Should see alternating slopes in the order of: feature +, feature -, conjunctive + then conjunctive absent for rt but in reality only diff is that absent higher than presentment. Wolfe says this is because FIT is just bottom up, but brain is involved too (top down)
Wolfe- guided search
Attentional engagement theory is search affected by how distinct the target is (how close the target looks to non targets) and how similar the rest of array is (e.g. if all the same, makes target pop out)- disputes FIT as say just colour difference should make target pop out but having a green dot amongst blue is quicker than light blue amongst dark blue
Guiding attributes
Core things that grab attention are colour, motion, orientation, size. Less so are luminance and curve, maybe are lustre and number but non are cateorgy, identity. Shading and lister grab attention which explains Instagram face (highlight and contour)
Wolfes model of guided search
Based on idea of top down processing e.g. quicker to find bread when in context of kitchen than in grid of other foods. Has selective pathways where info comes up, slow, id of stim elements, bottom up. Also has non selective pathway which takes gist of scene w semantic info and provides guidance and controls selective pathway attention- happens quickly
Values - Anderson 2011
Ps had to say orientation of line in both red and green circles but only rewarded for getting it correct in red- did many times to reinforce. In test, had to find unique shape amongst 5p6 others and either had red (high), green (low) or no distrscotr. High value increased rt but ind diffs like low wm and higher impulsivity increased rt for distract or
Tutorial
Simons- reasons for change blindness: initial scene is overwritten by changed scene, nothing is stored or only few details, nothing is compared (encode both but cog system doesn’t compare), feature combo (encode both to make new rep). Only encode first not second. Fenn et al- experiment series showed don’t notice change in voice unless attending - not due to DC, as ppl said no change even when told there would be, wm as task the same but different answer, remember more salient features e.g. gender change/ more central - take gist of environment . CHANGES IN VALUED OBJECTS MORE NOTICED
