Studying Memory Encoding with fMRI

Event-related vs. Blocked Designs

Aneta Kielar

Memory • Memory encoding: Process by which

experience of an event is transformed into memory trace.

• Main Questions:– Why some events are remembered and

others forgotten?– What brain regions/neural processes are

required for formation of durable memories?

Memory

• Remembered Events:– Undivided attention– Semantic vs. nonsemantic processing

• Brain Regions: ERP, fMRI, lesions studies

– Medial Temporal Lobe• Hippocampus & adjacent regions

(parahippocampal cortex)

– prefrontal cortex (modality dependent)

Memory• Problems

– Lesion studies• Encoding = Storage = Retrieval

– fMRI: blocked designs• Trials from each conditions presented sequentially,

inseparable from each other.• Allow comparison between encoding conditions• Do not allow direct trial- by- trial comparison between

encoding trials and subsequent memory

– ERP studies• Allow for trial–by- trial comparison• Limited spatial resolution

Memory

• Role of MTL in memory encoding = ?– MTL essential for memory encoding– MTL activation absent in some studies– Parahippocampal gyrus: memory encoding– Novel > familiar (Gabrielli et al., 1997)

– Novelty detection?

• It has been difficult to determine which regions underlie the process of memory encoding.

Memory and fMRI

• fMRI studies can distinguish between encoding and retrieval by measuring brain activation at each stage of memory

• Event-related designs: separate neural activations are recorded for each stimulus

• Allow direct comparison between specific encoding trials that lead to subsequent remembering and forgetting.

• Wagner et al. (1998) & Brewer et al (1998): whole brain fMRI to investigate whether neural activity in certain brain regions predicts later memory performance.

Building Memories: Wagner et al. (1998)

• Experiment 1: Blocked Design

– Effect of manipulation of encoding task (sem vs. nonsem) on prefrontal and MTL activation

Procedure:Participants: n =12

Encoding Condition (Scanning Session):Semantic processing: abstract vs. concrete (10 w each, 1 sec)Nonsemantic processing: upper of lower case

Alternating blocks: semantic, nonsemantic, visual fixation

Novelty of words was equivalent*

Procedure

Fix=8 msNS = 40sec

Fix=24 ms

Sem = 40sec

Fix=24 ms

NS = 40sec

Fix=24 ms

Sem = 40sec

X 4 scans+ + ++

Blocked Design

Blocked Design*: Segregate different cognitive processes into distinct time periods; many trials of the same type presented sequentially, trials

inseparable from each other during functional scan

Memory Test: 20-40 min later; yes/no recognition

Results• Behavioral:

– Decision RT• Semantic (873 ms) > nonsemantic (539 ms)

– Memory Accuracy• Semantic (85%)> nonsemantic (47%)

• Imaging Results:– Activation

• Word > fixation• Memory encoding, stimulus perception, response

generation

Results: Exp 1

A: LIFG; B: RIFG,

C: LPC

D:LIFG; E: Frontal Operculum; F:

LMTG

G: Visual cortex;

I: fusiform gyrus;

H: Parahippocampal gyrus

WORD > FIXATION

Results: Exp 1

• To identify regions of differential activation during encoding yielding high vs. poor memory

• Semantic vs. nonsemantic conditions– Semantic >nonsemantic

• Left prefrontal cortex• Left parahippocampal & fusiform gyri

Results: Exp 1

SEMANTIC > NONSEMANTIC

A: LIFGB: RIFG

D: LFG

E: bilateral frontal operculum

H: parahippocampal;

I: fusiform

• Blocked Design results: temporal and prefrontal brain processes influence encoding

• But they do not directly specify encoding differences that predict whether a specific word will be remembered or forgotten.

• Difficult to present blocks of forgotten and remembered items.

Experiment 2: Event-Related Design

• Event-related designs: associate brain processes with discrete events, which may occur at any point in the scanning session.

• Allow direct comparison between encoding trials that are remembered or forgotten

• Q: Does trial-by-trial diff. In encoding activation predict subsequent memory for experience?

Experiment 2: Procedure

• N =13 • Performed single incidental encoding task

• Measured neural activity elicited by single item; Direct comparison between encoding trials

• Encoding trials (Scanning Session)

– Semantic decision: abstract vs. concrete

– 40 abstract, 40 concrete word trials, 40 fixation trials pseudo-randomly intermixed

Procedure

• Memory test: 20 min later

• Recognition Test:– 480 studied & 480 unstudied – Confidence judgment: high vs. low, new

+C A+ C AA CX 6

750 ms ISI = 1250 ms

+ + + + + + + +

Results

• Behavioral:– High Confidence RT > Low Confidence RT > Misses– Accuracy:– High Con Hits = Low Con Hits = Misses

• fMRI Results:– Encoding trials categorized based on whether word was

remembered or forgotten on memory test

• Trial types:– High confidence hits; low confidence hits, misses,

fixations

Results

• Activation : Word > Fixation

A: LIFG; B: RIFG; C: LLPC D: anterior/ventral LIFG; E: frontal operculum; F: LMTG; G: visual cortex; fusiform gyrus; H: Parahippocampal gyrus

Results

• Identification of regions with differential activation during encoding of words remembered & forgotten.

• High confidence hits > misses– Left prefrontal regions– Left parahippocampal– Fusiform gyri

Frontal Regions Temporal Regions

More Results

• Effect of processing time: ?– Matched RTs for High confidence hits & missed– Activation for remembered items: left prefrontal

& temporal regions

• Accuracy: ?– Accuracy during encoding same for high

confidence hits & misses.

Summary of Results• Neural signature during encoding differs for

subsequently remembered & forgotten events.

• Engagement of left prefrontal & medial temporal regions during encoding makes experience memorable.

• Activation of left parahippocampal gyrus greater for events later remembered

• Role of parahippocampal gyrus extends beyond novelty detection; general encoding– All stimuli equally novel

Conclusions• Prefrontal cortex:

– Memory encoding– Role of semantic & phonological processing – Effective memory encoding: analysis of meaning of

the item.– Organize info in WM

• Parahippocampal Gyrus:– Novelty detection & more…– Respond to memorable items– Attentional processing (Rugg, 1998)

Conclusions/Questions

• Better memory depends on activation of prefrontal and temporal processes.

• No activation found in hippocampus proper?– What does it mean for memory theories?

• What is the nature of relation between prefrontal & parahippocampal processes?– Independent or serial processing?

Blocked vs. Event-related Designs

• Blocked Designs:• Trials from each conditions presented sequentially,

inseparable from each other.• Allow comparison between encoding conditions• Do not allow direct trial- by- trial comparison

between encoding trials and subsequent memory

• Event-related designs:• Separate neural activations are recorded for each

stimulus• Allow post-hoc trial sorting• Can define experimental conditions based on

participants responses

Making Memories: Brewer et al. (1998)

• Used event-related design to identify brain regions during encoding that differentiated between visual experiences that were later remembered or forgotten.

• Procedure:– n = 6

• Encoding (Scanning Session):– Subject viewed complex color photographs– 24 pictures of outdoor and indoor scenes– Judgment task: indoor or outdoor scene

Procedure

2.88 ms

OI OI I OO IX 4

ISI = 12.96 ms

++ ++++ + +

Procedure

• Memory Test:– 30 min after encoding task– Recognition test: 96 seen pictures & 32 new

pictures– Basis for recognition: remember or know

• Response types:– Remembered, familiar, forgotten

Results• Behavioral Results:

– Remembered: 25%– Familiar: 38%– Forgot: 48%

• fMRI Results:

• Activation Map– Pictures > fixations

ResultsScenes > Fixations

DLPF, Thalamus,

fusiform,

parahippocampus,

visual areas

Results• Computed correlations between size of

event- related response and memory classification of remembered, familiar and forgotten pictures.

• Regions where event related activity was correlated with greater memory:

– 6 in bilateral parahippocampal cortex– 1 in right dorsolateral prefrontal cortex

RDLPF

Bilateral Parahippocampal

Cortex

Activation during Encoding & Subsequent memory

Results• Determined source of the correlation between

event-related study phase activation and subsequent memory.

• Summed parahippocampal activation in each individual subject.

• Activation

• Remembered > Familiar

• Familiar > Forgotten

Summed parahippocampal activation

Results• Parahippocampal activation during study

predicted: – When picture would be remembered or

forgotten

– Predicted whether pictures would be more or less centrally remembered.

• Only frontal and parahippocampal activations predicted subsequent memory.

Conclusion

• Degree of Activation in Right Frontal & Bilateral Parahippocampal regions measures how well a particular visual experience is encoded and remembered.

Conclusions

• Interactions between Right Frontal WM processes and parahippocampal long term-memory processes

• right frontal & parahippocampal activation

encoding

memory