Prof. Greg Francis 5/27/081AttentionIIE 269: Cognitive PsychologyGreg FrancisLecture 13Should you pay $59.95 for Mega-speedreading?Attention We saw last time that attention can havevery powerful effects when it is focused on one thing, you ignoreother things Today we want to consider some morespecific properties of attention and look at experimental methods that areused to study attentionCharacteristics of attention By identifying the properties andcharacteristics of attention we can deduceproperties of the underlying systems thatare involved in cognition whether attention is thought of as a “system” or as a by-product of other systems Look at temporal featuralAttentional blink Suppose you have to identify rapidly presented (100ms) letters e.g., detect J and/or K in a stream of lettersMPKRWSAttentional blink Turns out that detection of first letter tends to makedetection of the second letter very difficult if it immediately follows the first Attentional blinkMPKRJSAttentional blink Measure frequency of detection get a curve like this (my CogLab data) compare to group dataImplies thatdetecting firstletter causes you to misssecond letter!Prof. Greg Francis 5/27/082Attentional blink Measure frequency of detection class data (72 observers)Implies thatdetecting firstletter causes you to misssecond letter!Attentional blink Suggests that processing the first targetletter, “J” uses up resources that would otherwise be usedto process second target letter, “K” attentional focus and refocus takes time and forthis task takes approximately 400 msSLNBJRKH...JOnce processing is done (afew hundred milliseconds),there is no difficultydetecting second letter.Attentional blink You could use the same type of approachto find the time course of attentionalprocessing with other stimuli shapes numbers colors faces Flowers spiders...Speed reading Ever seen the ads for speed reading? One is for a book called Mega Speed Reading claims to teach you to read 25,000 words per minute ==> 2.4 milliseconds for each word impossible, if only by attentional blink! The seller is skimming at best, and lying at worstVisual search Find the mistake in this sign Much of our time is spent looking for variousthings Staplers Road signsVisual search Time and type of processing can also be measured byhaving observers respond as quickly as possible whenthey detect a target hypothesize that tasks that involve attention will beslower than “perceptually” based tasks expect attention to depend on the number of things thatmust be searched Typically, we distinguish between a target anddistracters by one or more features we vary the number of distracters and measure reaction timeProf. Greg Francis 5/27/083Feature search When the target differs from all the distracters by onefeature (color, shape, size, orientation motion,…) which feature difference leads to fast detection of the target? does it depend on the number of distracters?Color Orientation ShapeConjunctive search When the target differs from all the distracters by onefeature (color, shape, size, orientation motion,…) but thedistracters also differ target is the conjunction (intersection) of features which feature differences lead to fast detection of the target? do they depend on the number of distracters?Color/shape Orientation/Color Shape/orientationImpressions Feature searches (at least for right features) effortless target “pops-out” at you reaction time does not vary with the number of distracters Conjunctive searches (at least for right features) effortful requires you to examine individual stimuli and determine ifthey are the target keep searching until you find the target and then hit theresponse key reaction time does vary with the number of distractersVisual search The data support these impressions with strong quantitative validation Four types of responses 1) Feature - present (can respond as soon as see target) 2) Feature - absent (must examine all stimuli before suretarget is not present) 3) Conjunctive - present (can respond as soon as seetarget) 4) Conjunctive - absent (must examine all stimuli beforesure target is not present)CogLab feature search Few distracters -easyCogLab feature search Many distracters - still easyProf. Greg Francis 5/27/084Conjunctive search Few distracters - fairly easyConjunctive search Many distracters - difficultVisual search My data from CogLab compare to group CogLab dataVisual search group CogLab data (69 observers)Visual search Interpretation Feature searches do not vary with number of distracters» feature searches “pop-out,” the distracters are irrelevant Feature present = feature absent» no searching is required, the reaction time just reflectsthe time to make the response Conjunctive searches increase RT with number of distracters» must search individual items and compare with target tosee if it matches, each comparison takes time (attention) Conjunctive absent > conjunctive present» must search through more items (on average twice asmany) for absent than present searches There are thousands of variations of visual searchexperimentsInterpretation Feature maps: color, shape Feature search can identify target within either feature mapShapeColorbluegreencornersarcsProf. Greg Francis 5/27/085Interpretation Feature maps: color, shape Feature search can identify target within either feature mapShapeColorbluegreencornersarcsInterpretation Feature maps: color, shape Conjunctive search cannot identify target within either feature map aloneShapeColorbluegreencornersarcsRequirescomparisonacross featuremapsSerial processthat takes timeInterpretation Feature maps: color, shape Conjunctive search cannot identify target within either feature mapShapeColorbluegreencornersarcsRequirescomparisonacross featuremapsSerial processthat takes timeAutomaticity When a task is unfamiliar it seems torequire a lot of attention to perform Later it requires less attention riding a bike driving a car typing tying shoelacesAutomaticity The process whereby a task goes fromrequiring a lot of attention to requiring little iscalled automatization Many tasks are automatizable color naming word naming Can measure effects by pitting
View Full Document