Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
*
Logic of our study2
Jarrod A. Lewis-Peacock1,2, Matthew R. Salesi3, Jonathan D. Cohen1,2, Kenneth A. Norman1,2
1Princeton Neuroscience Institute, 2Department of Psychology, 3Department of Computer Science, Princeton University
Behavioral performance4
“Prospective memory”1
Task & procedures3 MVPA: cross-validation (train test)5
Decoding the use of working memory & episodic memory in prospective remembering
Balance two simultaneous tasks
MVPA: trial-averaged decoding & behavior6
How do we remember to execute our plans at the appropriate time?
e.g., Think to yourself ... “I need to buy orange juice”
Working Memory active maintenance
Episodic Memory cue-based retrieval
(think aboutother stuff)
costly but reliable
cheap but unreliable
Bias strategy used for picture task
high proactive interference & easy lexical decisions (1-back)
Working Memory
low proactive interference & hard lexical decisions ( 2-back)
Episodic Memory (EMbias)
Assess fMRI for evidence of strategyUse multivariate pattern analysis (MVPA) to measure theactive, sustained retention of the target stimulus
Predictions:
MVPA readout should correlate with prospective memory performance more in WMbias trials than in EMbias trials
Episodic retrieval success effects in posterior parietalcortex should be greater in EMbias trials than in WMbias trials
1. Picture recognition2. Lexical decision (n-back)
(WMbias)
N=14 participants; 18-35 years old, healthy young adults
* Classification done separately for each subject, using L2-penalized logistic regression,in anatomically-derived voxel mask consisting of fusiform & parahippocampal gyri
Result - comparable performance in both conditions
Society for Neuroscience Annual Meeting, Washington, D.C., November 13-16, 2011
left Ang. Gyrus
Preliminary results suggesta dissociation of hits/misses
in left dlPFC (WMbias) & left Angular Gyrus (EMbias)
left dlPFC
* **
target2 s
mythbrain
jalp
probe12 s
delay2 s
probe22 s probe3
2 s
zrud
probeN2 s
N = {1, 2,..., 21}
feedback2 s iti
6 s
1-backlexical decisions
small set, repeatedpictures
pack
2-backlexical decisions
large set, trial-uniquepictures
tumbate
keace
WMbias
EMbias
scene(30)
-none-(30)
face(30)
WMbias WMbias WMbias EMbiasEMbiasEMbias
6 blocks (15 trials each), alternate condition (random 1st)
829.04
see: McDaniel & Einstein (2007); Reynolds, West, & Braver (2009); Gilbert (2011);
Lewis-Peacock, Drysdale, Oberauer, & Postle (in press)
Research funding provided by NIH grants R01 MH069456 & R01 MH075706
GLM7Result - classification is best when training &/or testing on WMbias trials
EM WM WM EM EM EMbias bias bias
WM WMbiasbias bias bias bias
WM WMbiasbias
WM EMbiasbias
bias
bias
bias
bias