Prof. Greg Francis 5/27/081Improving memoryIIE 269: Cognitive PsychologyGreg FrancisLecture 22How to improve your memory withoutspending $80.Memory We seem to be unable to control ourmemories learn things we don’t want to remember unable to learn things we want to remember Is there any reliable cue that something willbe remembered? no but there are several tricks you can use toimprove memory in certain situationsStudy style Generally, more study leads to bettermemory Style of study matters too distributed practice is better than massedpractice avoid cramming! true for many skills3 hours1 hour1 hour1 hourStudy style Consider encoding specificity to remember something in many contexts, studyin many contexts» dorm room, bus, library,... to remember something in single context, studyin same context» classroomLevel of processing Rehearsal is not as important as the depth ofprocessing Craik & Tulving (1975) Subjects observe words with associated tasksquestion YES NOIn capitalletters?BOOK bookRhyme withthing?spring sprintSynonymfor heavy?bulky brownLevel of processing Recall is better as depth of processingincreases0510152025Capital Rhyme SynonymStudy taskNumber of words recalledProf. Greg Francis 5/27/082Intention Level of processing is more important than intent tolearn Five groups of subjects Perform a “shallow” study task» (a) Intentional learning : told they will be tested to recall thewords» (b) Incidental learning: not told they will be tested Perform a “deep” study task» (a) Intentional learning : told they will be tested to recall thewords» (b) Incidental learning: not told they will be tested Control: told they will be tested to recall the words» not given any study taskIntention Shallow task was to look at letters in the word Deep task was to rate the pleasantness of the word Control group could do anything they wanted tomemorize the words they knew they would be testedDoes word have an E or a G in it?Rate the pleasantness of the word.RABBITshallowdeepIntention Recall improves with task, but does not varywith intention0510152025Shallow Deep NoneTask typeNumber of words recalledIncidentalIntentionalControlNotecontrol!Implications People do not know the best way toremember! There was nothing stopping controls from applying“deep” processing they apparently did not» or they would have done better! You will do best if you study interactively read notes rewrite notes rephrase notes teach someone elseMemory trick - grouping We often hear of people memorizingpages of the phone book how do they do it? some cheat (frauds) others take advantage of organization andmemory tricks SF learned to increase his digit span to81 digits (any random sequence) 200 hours of practiceSF: Digit span Broke down and organizedeach digit list Long-distance runner sequence like 3492 converted to“3 minutes 49.2 seconds- nearworld record time” Eventually created a hierarchyof tricks Technique did not transfer toother memory tasks (e.g.,letters)Prof. Greg Francis 5/27/083Method of loci Used by ancient Greeks to remember complicatedspeeches To remember a list of words or key ideas visualize walking around an area with distinctive landmarks link the items to be remembered with landmarks to recall items in order, mentally walk through area (any ordered sequence will work -- e.g., a children’s rhyme)Method of loci e.g., grocery listhot dogscat foodtomatoesdrivewaygarage interiorfront doorITEMS LOCIAdd vivid,bizarre imageryMethod of loci This technique works Subjects study a list of 25 words used method of loci or any method they wanted0204060801001 5Delay prior to recall (weeks)% words recalledLociControlMnemonists Some people seem to have extraordinarymemories professional - apply one of the techniques we’vediscussed spontaneous- seem to not consciously apply atechnique Photographic memory? few documented cases Not all with happy endingsS.: Luria Luria: Russian psychologist met S in 1920s S able to recall without error a list of 70 words» took 2-3 minutes» able to report it again several months later other unusual characteristicsS.: Luria Extreme synesthesia sensory information from one modality evokessensation in another tone, 30 cps, 100 decibles --> “saw” a strip 12-14cm wide the color of old, tarnished silver» 50 cps--> brown strip, taste of sweet and sourborscht voices gave rise to visual responses used the full sensation of events to help memoryProf. Greg Francis 5/27/084S.: Luria Visual imagery used method of loci such strong imagery it interfered with his ability tounderstand simple prose» words kept evoking inappropriate images...Sleep Many types of memory improves with sleep Some type of “consolidation” of memories The effect is not just time Although time also has an effect We’ll look at one representative study Ellenbogen et al. (2007)Sleep Subjects learn toidentify relationshipsbetween “random”shapes Only shown one pairat a time Subjects have tolearn/memorize theappropriate answer toeach pairSleep There is a ordered arrangement to the stimuli If you know this arrangement, deciding for any pair is easy But subjects are never explicitly told about this arrangementSleep Subjects are split in tothree groups, accordingto when they are tested 20 minutes later 12 hours later 24 hours later No differences whentested on the originallystudied itemsSleep Subjects are split in tothree groups, accordingto when they are tested 20 minutes later 12 hours later 24 hours later Big differences whentested on new pairs thatfit the ordered structure E.g., A>C, C>E, B>DProf. Greg Francis 5/27/085Sleep Half of the 12 hour grouphad sleep and half didnot It makes a difference forpairs of items that are farapart in the orderedstructure 1-degree: A>C, B>D,… 2-degree: A>D, B>E,… Lesson: study early! Get some sleep!Practicing recall Karpicke & Roediger (2008) Subjects study 40 Swahili - English word pairs mashua -- boat Test for English given Swahili:» mashua --??? Four groups of subjects, that differ after an item is correctlyrecalled ST (study-test): subject studies and is tested over every pair SnT (study on non-recalled - test on all): when a