4/27: Episodic memory: Retrieval.doc

4/27: Episodic memory: Retrieval E.P. Video Hippocampus Positioned to hear about the different things that are occurring in a moment in time for the organism MTL damage MTL is the site at which the different features of an event come together The site at which episodes are born MTL Bilateral damage Global amnesia = not a deficit for a particular kind of knowledge, but cuts across modalities, all classes of stimuli, all the ways of probing for memory Animal models Importance of spatial coding in the hippocampus Do you need to have extensive damage to the MTL to have an amnesic condition? No Oxygen deprivation can cause it Usually causes slight damage to hippocampi Hippocampi are really sensitive to oxygen deprivation Damage to a single subfield of the hippocampus Lesion was confirmed postmortem Relational Memory Theory Hippocampus is a convergence zone in which everything in neocortex projects into the hippocampus through cortical structures Conjunctive representations = capture conjunction of event elements that are occuring in a moment in time One piece of the conjunction can later cue another piece of the conjunction Tenets of Relational memory theory 1. Hippocampus supports declarative memory (memory for facts and events) 2. Declarative memory is a fundamentally relational system Codes for the relations between things 3. Hippocampus contributes to the critical process of relational memory binding and activation of conjunctions 4. Represents and retrieves all manner of relations among the elements of experiences, events, or scenes 5. Rapidly acquires these representations (1-trial learning) 6. Representations are flexibly addressable Can cue things with different pieces of information 7. Parahippocampal region contributes to memory for individual items Perirhinal and parahippocampal only hear about pieces of an event It's only the hippocampus that gets the full ensemble of pieces Dissociable MTL encoding effects: associative vs. item recognition Associative memory vs. item memory Ps see images of individuals and a name that goes with that individual Delay period Test phase Probe knowledge for pairs of items 1. Do they recognize it at all? 2. Did they encode the association between them? 3. Did they recognize and encode the association? Test probes Intact = knowing which name went with picture Recombined = recognizing, but not remembering the association New/forgotten = not recognizing Brain responses Hippocampus intact > new > recombined Perirhinal recombined > intact > new Consolidation Brain injury Full anterograde amnesia Partial retrograde amnesia Canonical view Retrograde amnesia isn't all-inclusive Temporally graded retrograde amnesia Things that were experienced closer to the injury are more likely to be lost than things in the distant past The fact that we experience retrograde amnesia only in period close to the injury is the main evidence for consolidation "Ribot's Law" = law of forgetting/law of regression New perishes before the old, the complex before the simple Standard model of consolidation MTL "pointer" and neocortical features traces constitute the memory You need to form conjunctive representation in the hippocampus It has to be able to get to cortical area There is a replay back out to cortex of the patterned event feautres This allows the cortex to gradually form connections across the event elements and encode the fact that they co-occurred No longer need the hippocampal representation to reactivate the full ensemble when you only have a cue What might be driving this cortical replay? Retrieval/remembering We retrieve memories in normal life Sleep Replays things that were present during the day Rodent study Listening in on 10 hippocampal neurons A given hippocampal neuron will fire when the rodent is in a particular place in space When you listen in during REM sleep, hippocampal cells have a structured response Above chance significant correlation of reinstating the pattern that was present when the organism was awake and having these events Hippocampus does some sort of replay during sleep of patterns that were present during awake period Replay during sleep period seems to be triggered by cues in cortex that trigger hippocampus that goes back to cortex and structures cortical structures (consolidation) Cueing sleep dependent consolidation Most amazing data point that he's seen Play concentration memory game Figure out where pairs of objects are during space Researchers waft odor into room Do it in the evening Then brought into MR scanner Wait for you to fall asleep and land in slow wave sleep Either present odor or a control odor Wake up next morning, surprise memory test no odor Results Odor during learning AND sleep: memory performance is 10% better than when control odor is presented during sleep No odor during learning, odor during sleep: doesn't affect performance Variation Odor during learning (whole time) Go into scanner while still awake Odor off and on while you're awake Fall asleep Slow wave Odor on and off while you're asleep What's the response in the brain when I cycle back in forth (odor/no odor)? Back out of the scanner Same benefits of sleep + odor No benefits of wake + odor slow wave sleep Enhanced response in hippocampus Something happens in hippocampus through time Engages in retrieval (replays cortical patterns) Sleep- structured responses Seem to relate to experiences the organism recently had Principles of episodic retrieval Cue-dependent nature of retrieval Schematic: probability that I'll be able to remember a past experience is heavily dependent upon the similarity of the cues that were present during encoding and cues present during retrieval Scuba divers study on land or in water recall either in the same place they studied or in the other location 2x2 design Context is crucial because memories are buried within their traces External context contains important stimuli Intoxication effects Alcoholics can struggle to remember important things when they lost things in an intoxicated state When they're drunk again, they find them State-dependent memory effect results are better if you studied drunk Mood effects Current mood/emotional state can bias what we remember Tried to nudge them to a more happy or depressed state when they're happy, memory is better for positive words when they're depressed, memory is better for negative words Important implications for mental health Principles governing retrieval Sampling theory Forgetting: our ability to remember depends on connections Spacing effect: cues change. the greater the lag, the higher probability that they're different Context changes Retrieval and global matching items (crazy equation) Free recall Context cue If there's very little competition of cues, you're likely to recall an event with only one cue Cued recall Getting multiple cues to drive retrieval Pattern completion