modeling individual differences in working memory capacity larry z. daily marsha c. lovett lynne m....

19
Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported by AFOSR grant F49620-97-1-0455 to Lynne M Reder

Upload: nathan-ward

Post on 18-Jan-2018

220 views

Category:

Documents


0 download

DESCRIPTION

Working Memory in ACT-R Limit on working memory is a limit on source activation This limit affects chunk activation Chunk activation affects the likelihood and speed of retrieval

TRANSCRIPT

Page 1: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

Modeling Individual Differences in Working

Memory Capacity

Larry Z. DailyMarsha C. LovettLynne M. Reder

Carnegie Mellon University

This work supported by AFOSR grantF49620-97-1-0455 to Lynne M Reder

Page 2: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

Working Memory

• provides the resources needed to retrieve and maintain information during cognitive processing

• as the working memory demands of a task increase, performance on the task decreases– Anderson & Jeffries (1985)– Anderson, Reder, & Lebiere (1996)– Burgess & Hitch (1992)– Salthouse (1992)

Page 3: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

Working Memory in ACT-R

• Limit on working memory is a limit on source activation

• This limit affects chunk activation

• Chunk activation affects the likelihood and speed of retrieval

W jwj constant

iA iB jw jisj

Page 4: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

Goals

• To continue the work of Lovett, Reder, & Lebiere (in press) and model working memory differences at the level of the individual subject.

• To further that work by showing that we can model subject performance at a fine grain

• To show that estimates of W from one task correlate with performance on a qualitatively different task

Page 5: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

The Oakhill Task

• Developed by Oakhill and her colleagues (e.g., Yuill, Oakhill, & Parkin, A., 1989)

• Modified span task– Subjects read all characters– Recall only digits

. . .

. . ....

..

....

..

.

a

f

. . . 8

. . .

. . ....

..

....

..

.

j

d

. . . 5

1st string

2nd string

<remaining strings>

Page 6: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

Model

• Chunks– Goals

• articulate• recall

– Memories• memory

• Productions– articulate

• read-aloud• create memory• rehearse-memory

– recall• recall-span• no-recall• read-item• next-item

Page 7: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

ArticulateProductions

READ-ALOUDIF the goal is to articulateand char is in visionand char has not been articulatedand char has an external representationTHEN say charand note that char has been articulated

CREATE-MEMORYIF the goal is to articulateand char is the last character of the stringand char has been articulatedTHEN create a memory of char in the current position on the current trialand move to the next position

Page 8: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

ArticulateProductions

REHEARSE-MEMORYIF the goal is to articulate on a trial

and the articulation has been doneand there’s a memory of an item in a

positionTHEN rehearse the item

and move to the next position

Page 9: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

RecallProductions

RECALL-SPANIF the goal is to recall a position on the current trial

and there’s a memory of an item in that position on this trial

and the item has not been recalledTHEN recall the item

NO-RECALLIF the goal is to recall

and there’s no memory of an item in thecurrent position

THEN recall blank

Page 10: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

Experiment 1: Aggregate Results

• Model parameters:– MP = 2.50– RT = 0.88– AN = 0.13– BLL = 0.50

• W fixed at 1.0• R2 = .99

BB

B

B

J

J

J

J

3 4 5 60.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Memory Set Size

B Data

J Model

Page 11: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

Experiment 1:Subject Data

B B B BJ J J J

3 4 5 60.00.10.20.30.40.50.60.70.80.91.0

Memory Set Size

B Data

J Model

Subject 100W = 1.5

BB

B

B

JJ

JJ

3 4 5 60.00.10.20.30.40.50.60.70.80.91.0

Memory Set Size

B Data

J Model

Subject 104W = 0.9

B B

B

B

JJ

JJ

3 4 5 60.00.10.20.30.40.50.60.70.80.91.0

Memory Set Size

B Data

J Model

Subject 105W = 1.0

B

B

B

B

J

J

JJ

3 4 5 60.00.10.20.30.40.50.60.70.80.91.0

Memory Set Size

B Data

J Model

Subject 107W = 0.8

Page 12: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

Experiment 1:Serial Position

B BB B B BJ J J J

JJ

1 2 3 4 5 60.00.10.20.30.40.50.60.70.80.91.0

Serial Position

B Data

J Model

Subject 100W = 1.5

B

B

B

B

B

BJ JJ

JJ

J

1 2 3 4 5 60.00.10.20.30.40.50.60.70.80.91.0

Serial Position

B Data

J Model

Subject 104W = 0.9

B B B

B

B BJ J

JJ J

J

1 2 3 4 5 60.00.10.20.30.40.50.60.70.80.91.0

Serial Position

B Data

J Model

Subject 105W = 1.0

B B

B B B

B

JJ

JJ J

J

1 2 3 4 5 60.00.10.20.30.40.50.60.70.80.91.0

Serial Position

B Data

J Model

Subject 107W = 0.8

Page 13: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

Experiment 2:Aggregate Data

B

B

BB

J

J

JJ

3 4 5 60.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Memory Set Size

B Data

J Model

•Zero free parameters •Model parameters:

– MP = 2.50– RT = 0.88– AN = 0.13– BLL = 0.50

•W fixed at 1.0 •R2 = .99

Page 14: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

Experiment 2:Subject Data

B

BB B

J

JJ J

3 4 5 60.00.10.20.30.40.50.60.70.80.91.0

Memory Set Size

B Data

J Model

Subject 201W = 0.7

B

BB

B

J

J

JJ

3 4 5 60.00.10.20.30.40.50.60.70.80.91.0

Memory Set Size

B Data

J Model

Subject 211W = 1.0

B

B

BB

J

J

JJ

3 4 5 60.00.10.20.30.40.50.60.70.80.91.0

Memory Set Size

B Data

J Model

Subject 216W = 1.2

B

B

B B

J

J

J J3 4 5 6

0.00.10.20.30.40.50.60.70.80.91.0

Memory Set Size

B Data

J Model

Subject 241W = 0.8

Page 15: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

CAM Sub-test• Pencil & paper adaptation of CAM battery

item (Kyllonen, 1993, 1994, 1995)

• 9 items of varying difficulty

• Scores on original version correlate with performance on a WM dependent task (Reder & Schunn, in press)

• Example item:

Page 16: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

W / CAM Correlation

• Estimates of W were strongly correlated with CAM scores› r = .55› r2 = .3025

› n = 29› W varied from 0.6 to 1.6› CAM varied from 3 to 9

Page 17: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

Conclusions

• Varying W captures individual differences in performance on a WM task

• Correlation with CAM supports the use of W as a model for WM capacity

• ACT-R can accurately model performance at the individual subject level

Page 18: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

(p read-aloud=goal>

isa articulatevision =charstatus nil

=char>isa characterexternal =string

==>!output! (“Saying ~A” =string)=goal>

status done)

(p create-memory=goal>

isa articulatetrial =trialvision =charflag laststatus doneposition =position

==>!output! (Memorizing ~A in position ~A

incremented to ~A” =char =position =next)=memory>

isa memoryitem = chartrial =trialposition =positionrecalled not

=goal>vision nilposition =nextrehearse =position)

Page 19: Modeling Individual Differences in Working Memory Capacity Larry Z. Daily Marsha C. Lovett Lynne M. Reder Carnegie Mellon University This work supported

(p rehearse-memory=goal>

isa articulatetrial =trialflag laststatus donerehearse =position

=memory>isa memorytrial =trialitem =charposition =positionrecalled not

=position>isa positionprevious =next

==>!output! (“Rehearsing ~A in ~A =char =position)=goal>

rehearse =next)

(p recall-span=goal>

isa recalltrial =trialitem nilposition =position

=memoryisa memoryitem = itemtrial =trialposition =positionrecalled not

==>!output! (“Recalling memory position ~A” =position)!eval! (push-last =item *answers*)=memory>

recalled done=goal>

item =item)