140 likes | 251 Views
Sleep-associated changes in the mental representation of spoken words ( Dumay & Gaskell, 2007). Alison Trude PSYC 525 15 September 2010. Lexical Memory. How do we know when a new lexical entry has been integrated into memory? Competition when compatible with input. +.
E N D
Sleep-associated changes in the mental representation of spoken words (Dumay & Gaskell, 2007) Alison Trude PSYC 525 15 September 2010
Lexical Memory • How do we know when a new lexical entry has been integrated into memory? • Competition when compatible with input +
Fictitious Word Learning (Gaskell & Dumay, 2003) • Participants heard nonwords w/ high overlap w/ real words (e.g., cathedruke) • Good performance on 2-Alternative Forced Choice Task • Indicates good phonological learning- can recognize what they heard A B cathedruce cathedruke
Fictitious Word Learning (Gaskell & Dumay, 2003) • Then, participants completed a Pause Detection task to check for lexical competition • Previous research showed that words w/ more lexical competitors before the pause had longer PD times • Effect of learning “cathedruke” only seen at retest (7 days later) Did you hear a pause? Y N cathedr_al
The Lexicalization Delay • Subsequent studies showed that lexicalization effects took 24 hours to emerge • Sleep shown to play a role in memory consolidation in many domains (e.g., procedural, perceptual, motor skills, some areas of speech) • During sleep, new information can be gradually transferred to LTM, thus preventing catastrophic interference
Design • 2 groups learned 24 novel wds; tested 12 and 24 hours later (PD, 2-AFC & free recall of novel wds) Retest Retest Exposure/Test PM Group 12 hours 12 hours AM Group
Some more details… • Participants learned 24 novel words that were made up of real words + addt’l syllable (e.g., “shadowks”) • Did 2-AFC task with these stimuli (“shadowks” vs. “shadowkt”) • Also did free recall of novel words • During PD, participants heard 48 real words- 24 that had a novel word competitor and 24 that did not. Heard in context consistent w/ competitor (“shadow_k”).
Results • Lexical competitor effect: • RTbaseline – RTcompetitor • So, a negative # means slower (more competition) for comp. than for baseline • At 12 hr, only PM group show effect, at 24 hour, both groups • AM group had slept by then *
Results • Free Recall task: • AM group did worse @ 12 hr, better after sleep • PM group improved @ both 12 and 24 hr
Discussion • Lexicalization of phonological forms is associated with sleep • Why did free recall get better with sleep? • Strengthening of episodic (phonological) information, or, perhaps lexicalization provides a boost • Compatible with models that use lexicalization delay to prevent overwriting
Why does sleep help? • Circadian rhythm: Sleep irrelevant, but something happens at a particular point in circadian cycle • Lack of input: little speech input during sleep, so less competition, better chance to consolidate • Sleep is essential: process that allows lexicalization to happen is specific to sleep • Some evidence that particular configurations of REM/non-REM sleep affect performance • Hippocampus may gradually feed new info into LTM (neocortical store) • Next step: associate stages of word learning with specific brain regions
Hot off the presses! • AMLaP 2010 • Magnitude of change in competition was correlated with count of sleep spindles (Tamminen et al.) • MEG study(Gagnepain et al.) • Posterior effect for consolidated words • Frontal effect for non-consolidated words • Time-locked to different parts of the stimuli
Lingering questions… • Ways to test whether sleep is necessary vs. other possible options (e.g., lack of input)? • Different types of memory needed at various levels of language processing/encoding? • Reasons for improvement on free recall task?