Download The Sternberg Task: Understanding Memory Experiments and Results and more Papers Psychology in PDF only on Docsity!
Housekeeping • Complete Sternberg Experiment by Thursda • Phonological similarity for next Thursday. • First experiment report due March 5. • Read Ch 4. The Sternberg Task PSY 373 Human MemoryFebruary 17, 2009
Choose one from this list to write about.supplemental readings. Note also trial-by-trial^ Experiments about which you mawish to write your first experimentreport 1. Modality effect2. Suffix effect3. Brown-Peterson4. Sternberg search (soon)5. Phonological similarity (next)
•^ “Think like a scientist”^ What makes a good experimentreport? •^ Understand^ what we did in the experimentwhy. •^ Understand^ the differences (or similarities)the classic experiment. •^ Discuss the data thoughtfully. •^ Discuss the relevant theory thoughtfully.
Sternberg, 1966 • Science paper. • post-Broadbent pre-Atkinson and Shiffrin Overview of today’s materia • The Sternberg Task.1. Sternberg’s classic finding2. Self-terminating serial exhaustive search3. Challenges to this model.
The phone company’s manycontributions to science (an asideprivate research funding) • Physics:^ Cosmic • Experiment done at Bell Labs.
background^ radiation, semiconductor, laser. • Computer science: C, Unix • Cognitive science:^
The Sternberg task, semantic indexing.
The Sternberg task • Let’s give it a try! Ma Bell’s interest in memory • People have to remember phone numbers. •... for a short time. •... and for a long time.
Recognition basics • Rather than^ produce^ the^ stimuli, Reaction time basics • Time takes to make a response. • Longer RTs presumably mean more thinking. • Longer RTs presumably mean worse mem • From looking at RTs we could infer somethingabout the cognitive processes that resultdecision.
we questions about “probes” • Probes are either “old” (from the list) or(not on the list). • Sometimes we’ll call old items “targets” anditems “lures.” • Correct responses are “yes” for old items“no” for new items.
Sternberg’s results • Near-perfect performance. • “Yes” slightly faster than “No” • Slower RTs for larger set size (s). • Linear increase, with slope of^38
±^2 ms/i
Sternberg’s results^ Sternberg, 1966
Sternberg’s theory: Serial exhaustivesearch • Serial,^ meaning^ one^ item^ at Sternberg’s results^ Sternberg, 1966
a^ time^ (easy generate linear relationship). • Exhaustive meaning that you don’t stop whenfind a match (necessary to explain same slopold/new). • Scanning should be at 25-30 Hz (!)
Any uncontrolled variables inSternberg’s procedure?^ How fast is 30 Hz? • Thirty times per second. • We can’t reliably perceive individual eventsthan about 10 Hz. • “Subvocal speech” estimated at 6-8 Hz for • 30 Hz is perceived as a tone!
Fixed set procedure produces neaidentical results^ Varied^ Fixed^ vs^ varied set • Varied: different digits each trial. (We did • Fixed set: Same digits each trial. • E.g. For the following trials, determine whether the probof the set^ 8 3 5 4. • How should things change with fixed set?
Fixed^ Implications of the Serial Exhau
Serial position effects in Sternberexperiments •^ Monsell (1978; see also McElree & Dosherfound recency (little primacy) when list prfast, tested at short delay. •^ Neath (1993) found serial position effecdepended on delay over very short time sc •^ Serial^ exhaustive^ scanning^ could Search model •^ Subconscious serial scanning. •^ Should be no serial position effects—youscan the whole list. •^ Using Sternberg’s procedure there aren’t S •^ Under other circumstances, there are.
not^ d these results.
... and also big serial position effect Monsell 1978 showed linear slope.
Is this primacy effect a consequence
Importance of Monsell, 1978 rehearsal?
-^ Not sufficient to show linear relationship tofor model of task. •^ If serial exhaustive search really used, onlysome circumstances. •^ Sternberg is^ emphatic
that there are no position effects in the Sternberg (1966) data. • Maybe speed of presentation big variable?delay?
Conclusions from Sternberg tas • When^ you^ have^ a^
set^ of^ items^ comfo “active” in memory for on the scale of seconds,a linear relationship with slope of 40 ms/itemobtained. • No one’s quite sure why this is (different theobut there are a number of things we’re quiteit’s not. • Could^ have^ something
to^ do^ with oscillations.
Segue
-^ In the Sternberg task, you have to activelyinformation in mind. •^ In “short-term store” or some other type ofavailable storage. •^ You then answer a question about the informationin memory, holding in mind a probe. •^ There are a great many tasks in which youto be able to hold and manipulate information.
Baddeley and Hitch model of womemory (4.1) •^ “Slave systems”—places to hold informadifferent types. •^ A^ “central^ executive” An example of using working me •^ Is Spain larger than Texas?1. Retrieve map of Spain.2. Retrieve map of Texas.3. Compare size of maps. •^ Map of Spain must be held in mind somuntil step 3. •^ Also need something to decide what inforto retrieve and coordinate steps.
to^ determine information goes where. • The “phonological loop” has received thattention. • Slave systems can retrieve information fro
Word-length effect results • Figure 4.2, 4.3. • But see Table 4.4. • What could^ go^ wrong
in^ word-length
The big problem with BaddeleyHitch’s model • We don’t understand immediate serial recall! • Long-term memory span (Fig 4.4) • “Memory Unchained” story experiments?
Assignment • Do the Sternberg Search experiment by Thursda • Experiment Report due March 5:^
start thin about what you want to do yours on. • Read Ch 4.
