Ada Leung PhD, Benson Ng BScDepartment of Occupational Therapy

University of Alberta

Faculty: Ada Leung & Benson NgRelationships with commercial interests: oNot Applicable

Outline of the talk Neural mechanism of auditory working memory Neuroplasticity resulting from auditory working

memory – One case Cross-modal plasticity – Two cases

Neural mechanism of auditory working memory

Working memory• Working memory refers to the ability to temporarily store

and manipulate information in a brief period of time.

• Dual-pathway model of auditory processing– Sound identity: ventral stream– Sound location: dorsal stream

Arnott et al., 2001

Auditory working memory1. Determine neural mechanisms of auditory working

memory

2. Any specific area(s) that is(are) responsible to location working memory load and category working memory load?

Participants 16 right handed, healthy young adults

9 women and 7 men

Mean age = 25.19 + 5.13 years

Normal hearing (pure-tone thresholds with normal limits for frequencies ranging from 250 and 8000 Hz)

Stimuli 0 degree

-90 degree (left)90 degree (right)

Humane.g.- Laugh- Cough- Cry

Musice.g.- Flute- Piano- Clarinet

Noisee.g.- Dog bark- Bird chirping- Door locking

• 1005ms duration, • digitally generated with a 16-bit resolution and a 12.21 kHz sampling rate,• passed through a digital-to-analog RP2 converter (Tucker-Davis Technology),• delivered at about 88dB via fMRI compatible headphone to suppress noise by 25 dB

Working memory task Location: 1-back, 2-back

RHuman

LNoise

LHuman

RHuman

LMusic

(1-back Target) (2-back Target)Instruction: L1 or L2 (10s)

Example of one block (20 stimuli)

Button press Button press

Working memory task Location: 1-back, 2-back

RHuman

LNoise

LHuman

RHuman

LMusic

(1-back Target) (2-back Target)

• Category: 1-back, 2-back

Instruction: L1 or L2 (10s)

Example of one block (20 stimuli)

RNoise

LNoise

LHuman

RMusic

RHuman

(1-back Target) (2-back Target)Instruction: C1 or C2

Example of one block (20 stimuli)

Button press Button press

Button pressButton press

Design (Pseudo-randomized)Block1 Block2 Block 3 Block 4 Block 5 Block 6

30s 30s 30s 30s 30s 30s

End of Scan464s(7m44s)

Begin scan

44s 40s 40s 40s 40s 40s 40s

•Four conditions:•location 1-back (L1)•location 2-back (L2)•category 1-back(C1) •category 2-back (C2)

•No. of blocks = 9

•Total no. of targets for each task = 45

Results - Behavioral

00.20.40.60.8

1

l1 l2 c1 c2

Hit

rate

00.020.040.060.08

0.10.12

l1 l2 c1 c2

Fals

e al

arm

rate

800850900950

100010501100

l1 l2 c1 c2

RT

for h

it ra

te

95010001050110011501200

l1 l2 c1 c2

RT

for f

a ra

te

Load: F(1,15)=56.30, p< 0.001 Load: F(1,15)=12.45, p< 0.005

Load: F(1,13)=5.71, p< 0.05Task: F(1,15)=11.32, p< 0.005Load: F(1,15)=11.83, p< 0.005

Results Main effects and interaction effects

All activations are significant at corrected p < 0.05 and > 196 µl.

Results Peak voxel activity at Right IPL

Discussion (Leung & Alain, 2011)

Frontal-parietal activation for auditory working memory

Two sub-regions in the parietal cortex are modulated differentially by working memory load:

Anterior ventromedial region – location load Posterior dorsolateral region – category load

Neuroplasticity resulting from auditory working memory training

Research questionsInvestigate the neural plastic changes in the “what”network after a course of auditory working memory training.

Training-induced neural activity on the “what”network – a case illustration

Mr. KB, male Age: 39 First stroke at March 19, 2011 Acute right subdural hematoma Multiple cortical infarcts, bilateral PCA, bilateral ACA.

Lesion involves right prefrontal and parietal cortices.

Cortical blindness Brain surgery: craniotomy

Demographic information Audiogram: less than 20 dB for 500 to 4000 Hz. WMS-II (Wechsler Memory Scale):

Letter-number sequencing: 5/21 Digit span forward: 9/16

Digit span backward: 6/14 Recognition memory (story): 15/30

MoCA (Montreal Cognitive Assessment): 10/22

The training 7 weeks of practice on auditory working memory tasks Stimuli are digits and letters Practice protocol:

5 days a week, 30 minutes a day. Only 1- and 2-back tasks in week 1. 1-, 2- and 3-back tasks in weeks 2 to 7 Total number of exposure to 1-, 2-, and 3- backs are the same.

Pre- and post- fMRI testing 9 blocks of 1- and 2-back each 20 stimuli per block Stimuli are digits and letters

Behavioral results - training

Behavioral results - fMRI

Hit Miss False alarm

1-back 37/46 (80%) 9/46 (20%) 1/134

2-back 30/46 (65%) 16/46 (35%) 6/134

Hit Miss False alarm

1-back 42/46 (91%) 4/46 (9%) 0/134

2-back 45/46 (98%) 1/46 (2%) 4/134

Pre-training scan

Post-training scan

Results

Results

Discussion (Leung et al., 2014)

Continuous practice of an auditory working memory task for 7 weeks as training induces:

Decrease of overall activation at the fronto-parietal network

Activities at adjacent regions are maintained Maintained neural activities at the precuneus

(posterior-lateral)

Activation decrease Increase in neural efficiency

Reflect sharpening of the response in a particular neural network so that fewer neurons now fire in response to a particular task or stimulus (Poldrack, 2000)

Typically observed in cognitive tasks N-back task (Hempel et al., 2004) Tower of London task (Beauchamp et al., 2003) Tests of verbal free recall (Andreasen et al., 1995)

Cross-modal plasticity after auditory working memory training

Client 1 a 38 year-old man

Right CVA involving the MCA

Restricted diffusion to the right parietal lobe, superior right frontal lobe and the anterior portion of the anterior medial frontal lobe

Enlargement of the lateral ventricles

19 years prior to the study

Client 1 Completed a course of rehabilitative program for

improving his motor skills

Did not receive any training for cognitive remediation

Completed an undergraduate degree and was working full time in a business firm

Continues to report cognitive deficits that are noticeable during work and home life

Client 2

a 37 year-old woman

Right CVA involving the MCA three years ago

had an episode of cerebral infarction in the brain-stem during her hospitalized for her stroke

Client 2 Completed a course of cognitive training in an

outpatient clinic

Completed an undergraduate degree and was working full time in a private company

Feels fatigue easily

Continues to have difficulties with memory and concentration that affect her performance at work

Distribution of n-back tasksWeek 1 Week 2 Week 3 Week 4 Week 5 Week 6

1-Back 13 9 7 6 4 2

2-Back 5 7 7 7 7 6

3-Back 2 4 6 7 9 12

Neuropsychological test scoresClient 1 Client 2

Pre Post Pre Post

RAVLT a

Learning (trial 5 – 1) 5 5 6 7

Interference 5 6 4 8

Trial 6 6 10 13 14

Delayed recall 5 9 13 14

50-word recognition 11 14 14 15

SDMT b

Accuracy (correct – errors) 46 45 35 39

CFQ c 37 44 46 36

a Rey Auditory Verbal Learning Test b Symbol Digit Modalities Test c Cognitive Failure Questionnaire

Performance during the training

Training effectsClient 1 Client 2

Auditory Visual Auditory Visual

1-b 2-b 1-b 2-b 1-b 2-b 1-b 2-b

Mean A1* 3.32 3.19 4.51 2.92 4.18 3.04 2.89 2.04

Mean A2* 3.43 3.05 3.64 4.24 4.23 3.90 4.03 3.88

Mean A2 – Mean A1* 0.11 -0.14 -0.87 1.32 0.05 0.86 1.14 1.84

SD A1* 0.42 0.62 0.85 0.49 0.79 0.11 0.74 0.48

Cohen’s d 0.27 -0.23 -1.02 2.68 0.06 7.90 1.54 3.83

Average Cohen’s d 0.02 0.83 3.98 2.69

Note: Mean A1 reflects the average of task performance in the pre-training testing and mean A2 reflects the average of task performance in the post-training measurement. Average Cohen’s d reflects the weighted mean of 1-back and 2-back task performance for the pre-training and post-training testing. 1-b = 1-back task; 2-b = 2-back task; all values represent d’.

fMRI resultsClient 1 Client 2

Discussion (Leung & Ng, 2015)

Client 1 showed minimal improvement on auditory and visual n-back tasks and no improvement on SDMT, and CFQ.

Client 2 demonstrated substantial improvement on all n-back tasks and behaviourial tests.

Client 2 substantial activation in the frontal-parietal network before

training, which subsided after training

extensive activations in multiple cortices including the occipital and frontal-parietal regions in the untrained visual tasks during the post-training testing

Discussion Cross-modal plasticity

General pattern: Activation decrease in frontal and parietal regions in the post-training

testing

Consistent with previous working memory training studies on healthy young adults (e.g., Garaven et al., 2000)

Increased neural efficiency and the use of more precise neuronal and functional circuits (Kelly & Garaven, 2005; Takeuchi et al., 2010)

Specific pattern:

Intact frontal-parietal network for processing working memory

Promoted a favourable condition for cross-modal transfer to take place, which increased the client’s ability to succeed in other cognitive tests and perception of cognitive gains

Thank you!