Experimental

NIH MH064498; NRSA GM007507 (UW-Madison NTP)

Transfer of working memory training observed in contralateral delay activityBornali Kundu, David W. Sutterer, Stephen M. Emrich, Bradley R. Postle

Neuroscience Training Program, Departments of Psychology and Psychiatry, University of Wisconsin - Madisonbornalikundu@gmail.com

Does intensive training on a working memory task transfer to behavioral and neural markers of

STM capacity?

IInnttrrooduducctiontion

Experimental designExperimental design

Training on working memory tasks improves performance on the task itself and results in changes in fMRI and DTI-based measures, whose effects local-ize to fronto-parietal brain regions - implicated in working memory and short-term memory (STM) performance (Olesen et al., 2004; Dahlin et al., 2008; Takeuchi et al., 2010).

Location STM taskLocation STM task

Experimental: Adaptive dual n-back Experimental: Adaptive dual n-back (n=15; based on Jaeggi et al., 2008; using Brain Work-shop http://brainworkshop.sourceforge.net/)

Psychometric Measures:Psychometric Measures:1. STM capacity (K value) derived from color-in-location STM task. 2. Raven’s Advanced Progressive Matrices score (RAPM; Raven, 1990)3. Operation Span score (OPAN; Turner & Engle, 1989)4. Stroop task performance (Stroop, 1935)5. Search efficiency from visual search task (Wolfe, 1994)

EEG measures:EEG measures:EEG recorded with a 256-channel EEG amplifier (Electrical Geodesic, Inc., Eugene, Oregon), sampled at 500 Hz and referenced to Cz. Electrode impedances were set below 50 kOhm. Offline, data were downsampled to 250 Hz, and re-referenced to average of both mastoid electrodes. Trials rejected based on eye movement artifacts. For search data, only target-absent trials used. Baseline taken over 100 ms prestimulus. CDA and CSA amplitudes determined from average over 300-800 ms window post-stimulus.

DelayCue

200 ms

400 ms200 ms Probe

2000 ms

ITI

3750 ms

Training

Pre-trainingcognitive

battery

Post-trainingcognitive

battery

100 ms 950 ms 2000 ms200 ms

DelayMemory

Array ProbeCue

Cue

200 ms

SearchArray

3000 ms

Color-in-Location STM taskColor-in-Location STM task

Visual Search taskVisual Search task

Training

Pre-trainingcognitive

battery

Post-trainingcognitive

battery

Control: Adaptive visuospatial Tetris Control: Adaptive visuospatial Tetris (n=15 ; http://www.gosu.pl/tetris/); no overt memory demands

1 hr/day, 5 days/wk, 5 weeks

ResultsResultsTrainingTraining

Working memory training transferred to STMWorking memory training transferred to STM. No significant training transfer to complex span (OSPAN), fluid intelligence (RAPM), interference control (Stroop task), or behavioral search effi-ciency (derived from visual search task).

Mea

n N

Lev

el (z

)

AverageTrainingGain

Experimental Task (dual n-back) Performance

Day of Training0 5 10 15 20 25−4

−3

−2

−1

0

1

2

3

fit y = A*x Sum

Tot

al T

etris

Sco

re (z

)

Control Task (Tetris) Performance

Day of Training0 5 10 15 20 25−2

−1

0

1

2

3

4

fit y=p1+p2*x

K

POSTPRE

ExptControl1.8

2

2.2

2.4

2.6

2.8

-b

Pre-Post Psychometric MeasuresPre-Post Psychometric Measures

−2

−1

0

1

2

Pre-training K

Ch

ang

e in

CD

A (µ

V)

1 2 3

Cha

nge

in K

(z)

Gain (z)−2 −1 0 1 2−2

−1

0

1

2

Change in CDA (z)

Cha

nge

in C

SA (z

)

P7/8P7/8

−1 0 1

−1

0

1

42

% A

ccur

acy

Pretraining ExptPosttraining ExptPretraining ControlPosttraining Control

Memory Load

0.65

0.75

0.85

Both groups improved with training. Learning on n-back task well fit by power law (mean adj R =0.81). Control group performance did not fit well to

power function (mean adj R =0.49), and thus was fit to a linear function (mean adj R =0.75).

2

2

2

ExperimentalControl

ExperimentalControl

Change in CDA amplitude correlatedwith pre-training K across groups

Change in K correlated with training gain for Experimental group

STM capacity (K) improved with training for Experimental group

VSTM performance improved withtraining for Experimental but not

Control group

r(5)=0.85, P<0.05

5

0

−5

5

0

−5

−200 0 200 400 600 0 200 400 600800 800

Contralateral Search Activity (CSA) for Target-Absent Trials

ms

PRE POST

Parallel far trans-fer of training to the CSA (derived from visual search, with no overt memory component) F(1,12)=4.84, P<0.05.

2

0

−2

Pre-trainingPost-training

ms

Pre-trainingPost-training

2

0

−2

−200 0 200 400 600 800

Con

trol

(µV)

Expe

rimen

tal

(µV)

Working memory training-related change in CDA correlated with training-related change in CSA

These findings suggest that transfer of working memory training effects may be strongest for tasks that engage the

same underlying neural networks.

P3P5

P7P4

P6P8

EXPT

CO

NTR

OL

CONTRAIPSICONTRACONTRA

PRE POST

0 200 400 600 800 0 200 400 600 800

3

0

−3

3

0

−3

Contralateral Delay Activity (CDA) for Load 4

Volta

ge (µ

V)Vo

ltage

(µV)

Pre-Post Event-Related Potential MeasuresPre-Post Event-Related Potential Measures

Pre-training

LOAD4LOAD2

0 200 400 600 8002

0

−2

ms

Post-training

2

0

−2

0 200 400 600 800ms

Neural marker of STM capacity (the CDA; Vogel and Machizawa, 2004), derived from color-in-location STM task, decreased with train-ing F(1,18)=6.92, P<0.05.

P3P5

P7P4

P6P8

2

ms

EXPT

CO

NTR

OL

Volta

ge (µ

V)Vo

ltage

(µV)

EXPT

CO

NTR

OL

Volta

ge (µ

V)Vo

ltage

(µV)

r(6)=0.36; n.s.

r(12)=0.65, P<0.05r(13)=0.30; n.s.

-0.4

-0.2

0.2

0.4

0.6

Ch

ang

e in

co

ntr

rala

tera

l w

ave

amp

(mic

roVo

lts)

ControlExperimental

WM training reduced contralateral differencewaveform amplitude (Load 4 versus 2)

for Experimental, not Control group

0

F(1,18)=2.78, P=0.11