1

Sleep States and Memory

Dr. Carlyle Smith

Dept. of Psychology

Trent University

Peterborough, Ontario

History and Background

• Two experimental designs have been usedextensively to study the relationship ofsleep to memory processes.

• 1) Learn Task - Observe Sleep - Retest• 2) Learn Task - Deprive Sleep - Retest

Animal Studies

• Training was done with rats in a 2-wayshuttle shock avoidance task.

• Animals were given 50 trials/day for twoconsecutive days.

• Sleep recording took place both beforeand for many days after the end oftraining

50 TRIALS PER DAY FOR 2 DAYS

MINUTES OF REM SLEEP RELATIVE TO BASELINE FOR LEARNING AND NON-LEARNING ANIMALS

Days

TD1 TD2 PTD1 PTD2 PTD3 PTD4 PTD5 PTD6 PTD7

REM

in Minutes from

Baseline

-10

0

10

20

30

Learning AnimalsNon-learning ControlsBaseline

Early Human Work

• In a study done in a practical setting, weexamined the changes in sleeparchitecture that might occur aftersubstantial learning in a real lifesituation.

• We examined the sleep of students beforeand after exams without typifyingmemory systems

REM Sleep Changes After Exams

• We sleep recorded 4th year HonoursPsychology students in the summer, whenthey were not taking classes of any kind.

• We recorded these same subjects 3-5days after their last Christmas exam inDec.

• We then recorded them again the nextsummer, when they again were nottaking courses of any kind.

2

Exam Study Con’t ….• Control subjects were same age

individuals that had been in Honours,and were now in tech or teachingassistant positions.

• Their intent was eventually to go on tograd school, but they were consolidatingdebts, waiting for a partner to finish, etc.

• They were sleep recorded in Summer 1,at Christmas and during Summer 2 aswell.

MEAN TOTAL NUMBER OF REMS FOR LEARNINGSTUDENTS AND FOR NON-LEARNING CONTROLS

Summer 1 Xmas Summer 2

Me

an

To

tal N

um

be

r of R

EM

s

200

400

600

800

1000Learning Controls

*** *

DISTRIBUTION OF REMS IN EACH REM PERIOD FOR LEARNING AND CONTROL SUBJECTS AFTER CHRISTMAS EXAMS

REM Period

1 2 3 4 5

Num

ber of RE

Ms (from

Baseline)

-100

-50

0

50

100

150

200

250

300

LearningControls

*

REMS DENSITIES FOR LEARNING AND NON-LEARNING CONTROL GROUPS FOLLOWING CHRISTMAS EXAMS FOR THE LEARNING GROUP

REM Period

1 2 3 4 5

Mean

RE

M D

ensities (fro

m b

aseline)

-2

0

2

4

6

8LearningControls *

Appetitive Bar Press Task

• We examined (Smith&Wong, 1991)changes in sleep after extended training inSkinner box with increasing levels ofdifficulty

• 1 Magazine (M), 3 Simple Bar press (A),FR -10 sched (2 days) and then 3 days witha difficult sequential operant (SO).

Con’t...

• There were two bars to press.

• During the Sequential Operant, reward wasonly given when rats could “move” the lighton a 3 x 3 matrix from upper left to lowerright - after which they got a pellet

• This required alternate bar presses or 2 leftbar, 2 right bar.

• Deviations reset the program.

3

Con’t...

• Only some rats learned the task. We calledthem Fast Learning

• Some rats could not learn and extinguished.Called Slow Learners

• Non-learning rats were not trained.

• FRC controls pressed at the FR-10 levelinstead of changing to SO task

Minutes of REM Sleep For Each of the Groupsfrom Pre - Training Mean Baseline REM

BAR PRESS CONDITION

M A1 A2 A3 FR1 FR2 SO1 SO2 SO3

RE

M S

leep in minutes

-10

0

10

20

30

FL - fast learningFRC - FR ControlSL - Slow learningNL - Did not learnMean Baseline REM

Human Studies

• The progress in using human subjectshas paralleled the progress in theunderstanding of memory itself.

• Different kinds of memory are nowbelieved to exist.

• The two main types of memory are -– 1) declarative - often explicit

– 2) procedural - often implicit

Declarative Memory and Sleep

• Declarative learning does not seem to berelated to any stage of sleep.

• Sleep deprivation following acquisition ofa paired associates task did not impairmemory for this task 1 week later

• Sleep deprivation did not impair recallfor of a set of memorizedneuroanatomical structures.

0102030405060708090

RE

MD

Stag

e 2

Aw

Con

trol

s(L

)

Con

trol

s(H

)

TrainingRe-Test

Effect of Sleep Deprivation on Memoryfor Neuroanatomical Structures

No differences% Correct

Cognitive Procedural Learning

• Cognitive procedural learning is one typeof procedural learning.

• This type of learning requires the learnerto develop a new cognitive strategy inorder to solve the problem (e.g. Tower ofHanoi).

4

Cognitive Procedural Memoryand Sleep

• Cognitive procedural tasks appear torequire REM sleep for most efficientlearning

• Sleep following acquisition of the Towerof Hanoi and Mirror Trace tasks shows amarked increase in number of REMs andREM density.

• There is no change in the amount ofREM sleep (minutes) or any other stage.

0

200

400

600

800

1000

1200

1400

1600

1800

PRE POST

TESTCONTROL

Total Number of REMs

REMs

*p < .0001

Night 2 Night 3

0

2

4

6

8

10

12

14

16

18

PRE POST

TestControl

*

Night 2 Night 3

REM DENSITIES

p < .0002

5

0102030405060708090

100110120

PRE POST

TESTCONTROL

Total REM Sleep

Minutes

Night 2 Night 3

No Differences

0

5

10

15

20

25

PRE POST

TESTCONTROL

% REM Sleep

%

Night 2 Night 3

No Differences

IQ Groups

• The test groups were separated in termsof High (HiQ; X=125.8), Medium (MedQ;X=114.7) and Low IQ (LoQ = 102.5).

• Controls were a mix of all levels @ X=108.3 (n=6/group, N=24)

• The higher the IQ score, the better werethe learning scores for the tasks.

IQ Con’t...

• The higher the IQ score, the greater wasthe number of post training REMs andthe larger was the REM Density measure.(HiQ>MedQ>LoQ).

• The intensity of REM sleep was alsocorrelated with success in taskacquisition and re-test.

TOTAL NUMBER OF REMS BEFORE AND AFTER TRAINING ON THE TOWER OF HANOI AND MIRROR TRACE TASKS

Nights

Baseline Post-training

Mean

Nu

mb

er of R

EM

s

0

500

1000

1500

2000

2500

LoQMedQHiQControls

***

**

IQ Con’t...• There was no correlation between

“native” REM intensity and IQ or taskscores.

• The largest REM intensity responseswere exhibited by Ss with the highest IQand the highest acquisition and re-testscores.

• Post-training REM intensity may be abiological marker for learning potential

6

NUMBER OF RAPID EYE MOVEMENTS FOR COMBINED TRAINED AND NON-TRAINED CONTROLS AT EACH REM PERIOD

REM PERIOD

1 2 3 4

Num

ber of Rapid Eye M

ovements

100

200

300

400

500

600

700

800

Combined TrainedNon -Trained Controls

PERCENTAGE REM FOR TEST AND CONTROLSAS A FUNCTION OF NIGHT

NIGHT 1 AND NIGHT 2

PE

RC

EN

TA

GE

RE

M

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30 TESTCONTROLS

Selective REM Deprivaiton

• Selective REM sleep deprivationfollowing acquisition of the Tower ofHanoi or a complex Logic Task results inmarked memory loss on re-test one weeklater.

• NREM awakened controls were notimpaired on these tasks.

Tower of Hanoi TaskImprovement after sleep deprivation

Groups

Controls (HC)

TSD REMD Controls (R)

NREM Awakening

Savings S

coresP

re - Post (# of M

oves to Solution)

-10

-5

0

5

10

15

20

25

**

*

Alcohol Ingestion

• One common type of behaviour oncampus is to work hard during the day,drink hard in the evening and then fallinto bed.

• Alcohol is know to induce a reduction innumber and density of REMs wheningested in moderate amounts (as few as3 drinks) just prior to bedtime.

Alcohol Con’t...

• After an acclimatization night in the sleeplab., we trained students in both adeclarative (Paired Associates) and acognitive procedural (Logic) task(Alcohol Night 1) .

• They then drank 3-5 drinks of vodka(1 1/4 oz/drink) and orange juice andthen went to bed (0.7g/Kg.).

7

Alcohol con’t...• Another group drank orange juice alone

after learning (OJ Night 1)

• A third Control group drank nothing (CNight 1).

• All groups slept a second night with nodrinking.

• On Night 3, the OJ group of Night 1drank alcohol, while the alcohol group ofNight 1 drank OJ.

% CORRECT PAIRED ASSOCIATE RESPONSES BEFORE AND AFTER ALCOHOL CONSUMPTION

Before and After Alcohol Consumption

PRE POST

% C

orrect Paired Associates

10

20

30

40

50

60

70

80

90

100

ALN1ALN3Controls

% CORRECT WFFs BEFORE AND AFTER ALCOHOL CONSUMPTION

PRE POST

% C

orre

ct W

ffs C

on

stru

cte

d

20

40

60

80

100

ALN1ALN3Controls

**

Experiment 2

• Was this an alcohol or a sleep effect?

• Trained subjects on Tower of Hanoi andPursuit Rotor at 4 PM in the afternoon.

• One group (Aft - Alcohol) then drankimmediately (0.9g/Kg) after learning andwent to bed 5 hours later.

• One group drank just before bed (Eve -Alcohol).

NUMBER OF MOVES TO SOLUTION ON THE TOWER OF HANOI TASK

PRE POST

Num

ber of Moves to S

olution

0

20

40

60

80

100Evening - AlcoholAfternoon - Alcohol

*

Pursuit Rotor Target Times for AFT- EtOH and EVE- EtOH Groups Before and After Alcohol Ingestion

Alcohol Condition

PRE POST

Mean N

umber of S

econds on Target

60

80

100

120

140

160

EVE-EtOHAFT- EtOH

*

8

Sleep Effects

• REM sleep was similarly modified inboth experiments.

• Both number of minutes of REM sleepand number of REMs were reduced inthe first half of the night in study 1.

• The density of REMs was reduced in thefirst half of the night in study 2.

Conclusions

• Even moderate alcohol ingestion prior tobedtime can modify sleep architecture,particularly REM sleep in the first half ofthe night.

• Moderate alcohol ingestion at bedtimecan induce memory loss the night oflearning AND/OR 2 nights after theoriginal learning has taken place.

Conclusions con’t...

• Interference with normal REM sleepeither the same night, or two nights afteracquisition of procedural material canresult in memory deficits of 20-40%.

• Alcohol at bedtime can impair memoryfor both cognitive procedural and motorskills tasks.

Conclusions Con’t ...• The REMs results implicate the PGO

system of the pons in efficient memoryconsolidation

• Animal studies (Datta et. al., 2000)support this theoretical position, sincethey have observed increases in P wave(PGO) activity in the rat followingsuccessful acquisition of the 2-way shuttleavoidance task