technique for the analysis of evoked and background eeg activity applied to young and elderly...

102
1 Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects A thesis submitted for the degree of “Doctor of PhilosophyBy Craig Goodman Submitted to the Senate of the Hebrew University October 2002 Jerusalem

Upload: craig-goodman-phd

Post on 16-Apr-2017

111 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

1

Technique for the Analysis of Evoked and Background EEG

Activity applied to Young and Elderly Subjects

A thesis submitted for the degree of

“Doctor of Philosophy”

By Craig Goodman

Submitted to the Senate of the Hebrew University

October 2002 Jerusalem

Page 2: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

2

EEG-שיטות חדישות לאנליזה של מעוררים פוטנציאלים ופעילות רקע של ה

בנבדקים צעירים ומבוגרים

חיבור לשם קבלת תואר

"דוקטור לפילוסופיה"

מאת

קרייג גודמן

הוגש לסינט האוניברסיטה העברית בירושלים

2002תשרי תשס"ג אוקטובר

Page 3: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

3

ה בהדרכתו של פרופסור חיים סומרעבודה זו נעשת

Page 4: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

4

This work was carried out under the supervision of

Professor Haim Sohmer

Page 5: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

5

תוכן הענינים

עמוד קיצורים וראשי תיבות תקציר באנגלית 1-19 מבוא 19 מטרות המחקר 19 חשיבות המחקר

ותשיט 19 19 נבדקים 19 גירויים לך הניסוימכשור ומה 21 22 עיבוד נתונים 29-55 תוצאות

תוצאות סיכום 55 55 דיון 62 זמני חביון N1 62

P1 64

N2 65 67 מידות של המשרעות 71 רשימת סיפרות תקציר בעברית

Page 6: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

6

CONTENTS

Pages

Abbreviation

Abstract in English

Introduction 1-19

Goals 19

Importance 19

Methods 19

Subjects 19

Stimuli 19

Apparatus and Procedure 21

Data Processing 22

Results 29

Differentiation between background and evoked activity – Time

distributions and amplitudes of the EEG deflection: Criteria for

determination of the evoked response applied to young and elderly

subjects 29

Onset, termination and duration of the overall evoked response in young

and elderly subjects 36

Time distributions of the deflections during background activity

in young and elderly subjects – Rate of deflections 40

Amplitudes of the deflections during background activity in young and

and elderly subjects 42

Relation between mean rate of EEG deflections during the 300 ms

pre-stimulus background activity and during the 300 ms overall

evoked post-stimulus activity 43

Time distributions of deflections related to the individual

components of the evoked response – Partial time distributions of

deflections corresponding to the components of the evoked response:

Page 7: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

7

determination and parameters used 43

Borders and duration (width) of the partial time distributions related

to the N1, P1 and N2 components of VEP in young and elderly

subjects 44

Mean number of deflections per single stimulus trial for the partial

time distributions of N1, P1, N2 in young and elderly subjects 45

Variability of latencies of the EEG deflections contributing to the

N1, P1, and N2 components of VEP 47

Time locking of the EEG response deflections 49

Comparison between latencies of the peak D-values of the partial

time distributions of deflections and latencies of conventionally

averaged VEPs 49

Amplitude amplification of the deflections contributing to the

evoked response components 50

Summary of Results 55

Discussion 55

Latency Measures 62

N1 62

P1 64

N2 65

Amplitude Measures 67

References 71

Abstract in Hebrew

Page 8: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

8

תקציר

בנבדקים EGE-שיטות חדישות לאנליזה של מעוררים פוטנציאלים ופעילות רקע של ה

צעירים ומבוגרים

רקע:

האנליזה של הפעילות החשמלית הנרשמת מהראש, גם פעילות רקע

Electroencephalography (EEG וגם פעילות מעוררת בתגובה לגירויים תחושתיים )

Sensory evoked potentials(EP היא בעייתית בגלל הקושי לזהות את מרכיבי התגובה ,)

לעומת EEG-לגירוי בתוך "רעש" הרקע. זה נובע בעיקר מהמשרעת הגדולה של ה

טן. בנוסף ק signal to noise ratio. זה מביא ליחס אות לרעש EP-המשרעת הנמוכה של ה

גישה ה. EP-( והמשרעת של מרכיבי הlatencyלה של זמני ההופעה )יש שונות גדו

ירויים, דהיינו גהיא לעשות מיצוע של הרישומים להרבה EP-המקובלת יותר לחקור את ה

Averaged EPאבל שיטה זו מבטלת את השונות הרגעית של ה .-EP על פני הזמן )שינויים

ו מהווה מידע חשוב לגבי מן. השונות הז( ובמשרעת התגובה מזמן לזjitter –בזמן ההופעה

תפקוד המוח ולא כדאי לאבד מידע זה על ידי מיצוע. שיטה נוספת כרוכה באנליזה של הרכב

. גם בשיטה זו יש התעלמות EP-ושל ה EEG-של ה frequency analysisהתדרים

מהשינויים הרגעיים.על מנת להתגבר על הבעיות האלו, פיתחנו שיטות חדישות.

ות:שיט

-השיטות החדישות מבוססות על תיאור סטטיסטי של הזמנים של הגלים ברישום ושל גודלם

שבתוכם התגובות -וגם אלה שלאחר הגירוי EEG-ה -הרקע –גם אלה שלפני הגירוי

-30אנשים נורמליים צעירים )גילאים 14לגירויים. יישמנו את השיטות האלה על קבוצה של

Reversingמסך שח המתחלף -שנים(. הגירוי היה ראיתי 65 – 80זקנים ) 14 -שנה( ו 18

checkerboard הגלים –. ערכנו אנליזה של המרכיביםN1, P1, N2 בתוך חלון הזמן עד

Page 9: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

9

א"ש לפני 300( ואת התקופה EP-אלפיות שניה )א"ש( אחרי הגירוי )המכיל את ה 1000

אלקטרודות על פני 16רקע(. הפעילות החשמלית נרשמה מסדרה של EEGהגירוי )

הרץ. זמן ההופעה והמשרעת של כל גל בסדרת הגלים לפני 30-1הקרקפת עם סינון בין

(, זמן ומשרעת. מזמני כל הגלים -ואחרי מתן כל גירוי נקבע ולכן לכל גל היה קוטביות ) + או

Time distribution of theגירויים הרכבנו "פיזור הזמנים של הגלים" 300בסדרה של

deflections בנפרד בשביל הגלים החיובים ובנפרד מהגלים השליליים. בנוסף, מתוך

Amplitudeמשרעות הגלים )החיוביים בנפרד מהשליליים(, הרכבנו "מאפייני המשרעות"

Profiles .המשך האנליזה התבסס על "פיזורים" "ומאפיינים" הללו .

התוצאות:

ותר בזקנים מאשר בצעירים. משרעות גלי הרקע קצב הגלים בזמן פעילות הרקע היה גבוה י

באלקטרודות האחוריות היו גבוהות יותר בצעירים מאשר בזקנים. בזמנים אחרי מתן הגירוי

קצב הגלים בזקנים היה גבוה יותר בהשוואה לזמנים שלפני הגירוי )פעילות רקע(. לא היה

הבדל כזה בקצב בנבדקים הצעירים.

)מעל קליפת המוח הראיתית הראשונית( בצעירים Ozת הפעילות המעוררת באלקטרוד

ראה נא"ש שלא 70 -התחילה מוקדם יותר אחרי הגירוי מאשר בזקנים. זה נובע מקיום גל ב

N1, .P1בזקנים. היה שונות גדולה יותר בזקנים בפיזור זמני הגלים בזמן מרכיבי התגובה

נים. לעומתם, בצעירים היתה מידה וגם סטיות התקן של הזמנים הללו היו גדולים יותר בזק

גדולה יותר של "נעילת זמני הגלים" לזמן מתן הגירוי. משרעות הגלים בזמן מרכיבי

היו גדולות יותר בצעירים מאשר בזקנים. ממוצעת המשרעות של הגלים היו N1, P1התגובה

בשיטת המיצוע EP–שהתקבלו ב N1, P1תמיד גדולות יותר מאשר המשרעת של המרכיבים

Averaged EP.

. בכל מקרה משרעת הגלים בזמנים EP–זה מעיד על ההפרעה ששיטת המיצוע מכניסה ל

שלאחר הגירוי היתה גדולה יותר מאשר משרעות הגלים זמן פעילות רקע, זה מעיד שיש

הגברה של משרעות הרקע כתוצאה ממתן גירוי. בנוסף, לא כל גירוי גרם להופעת גלי תגובה

חלק את הפעילות המתעוררת בתגובה לסדרה של גירויים לשתי תת קבוצות: ולכן ניתן היה ל

Page 10: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

10

וקבוצת גירויים שלא גרמה לתגובה. ניתן Ozגירויים שגרמו לפעילות מעוררת באלקטרודת

לערוך אנליזה נפרדת לשתי קבוצות הגירויים הללו.

מסקנות:

בפרט יותר בקשר שיטות האנליזה החדישות מספקות הרבה יותר מידע באופן כללי, ו

להבדלים בפעילות החשמלית של המוח בין צעירים לבין זקנים. מידע כזה לא ניתן להשיג

בשיטת המיצוע המקובלת. השיטות מצליחות גם להפריד בצורה מתמטית סטטיסטית בין

רמו ת, לזהות במדויק את זמני תחילת התגובות וסיומן, ואיזה גירויים EP–לבין ה EEG–ה

נוסף, התוצאות נותנות הבנה טובה יותר בקשר למנגנונים המוחיים הגורמים תגובות. ב

והגברה. ניתן אפילו לשפר את Time locking -דהיינו "נעילת זמן" EP–להיווצרות של ה

על ידי מיצוע של התגובות לאותם הגירויים שגורמים לתגובה. Averaged EPשיטת המיצוע

תוארו כאן תאפשרנה לחוקרים לקבל מידע רחב יותר לגבי לסיכום, סביר להניח שהשיטות ש

מספר רב של קבוצות מקרים עם פתולוגיות שונות בפיזיולוגיים -ההבדלים הפיזיולוגיים והפתו

מידע שלא ניתן להשיג בשיטות המקובלות. -

Abstract

Page 11: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

11

Technique for the Analysis of Evoked and Background EEG Activity applied to

Young and Elderly Subjects

Background – The analysis of scalp-recorded brain activity, both ongoing EEG

and in response to sensory stimuli, has continuously proven to be problematic for

researchers in terms of the identification and analysis of components in response to

sensory stimulus. This is due to the relatively high level of background EEG activity,

providing a low signal-to-noise ratio (SNR), as well as the high variability of the

single response trials in terms of their latencies and amplitudes. The most common

approach to the study of EEG activity in response to repeated stimuli is to use

averaged evoked potentials (EPs). This method however contributes to the loss

valuable information, such as moment-to-moment variation with respect to time (i.e.

jitter), and similarly with respect to amplitude. It is likely that these variations contain

important information about brain function, and should not be ignored when

investigating different groups or pathologies. An additional technique is based on

frequency analysis of the EEG and EP. However, this too can be problematic in

regards to cortical brain activity at specific times, since the time-integrative aspect of

this form of analysis does not address trial-to-trial variations, much like the technique

used for conventional averaging. In order to overcome the problems generated by

conventional evoked potential averaging, new techniques have been developed in this

laboratory.

Methods - The new techniques are a more simplified approach to the analysis of

single response sweeps, which was developed in the hope of providing a more

practical and less cumbersome alternative, without sacrificing important information

regarding the electrical activity of the cortex. The techniques are based on statistical

descriptions of the times and amplitudes of the EEG deflections recorded before

(background) and after (evoked) a series of stimuli. These techniques were applied to

Page 12: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

12

young (18-30 years) and elderly (65-80 years) normal subjects, using a reverse

checkerboard visual stimulus. VEP components (N1, P1, N2), were analyzed in young

and elderly subjects using these new techniques. The window of analysis was 300 ms

before and 1000 ms after each stimulus trigger, and the frequency content of the

recording was limited by a bandpass filter of 1-30 Hz (with additional smoothing by

an algorithm). A single trial (EEG recorded before and after each stimulus) was

defined as a sequence of events (deflections), each having a time coordinate (with

respect to the stimulus), polarity and amplitude. These trials were quantified by

computer algorithms, which detected all positive and negative deflections throughout

the recording. The times of the positive and negative deflections were calculated for

all sweeps (stimulus trials) in a single session (n=300 trials) for the given subject.

Time distributions of deflections (separately for positive and negative) termed Time

distributions of deflections as well as separate distributions of the amplitudes of the

deflections termed Amplitude Profiles were constructed as a function of time with bin

widths of 6 ms. The average of six different recording sessions from the same subject

was calculated.

Results - A higher rate of background activity in the elderly subjects when

compared to the young across all 16 electrode sites. The amplitudes of the deflections

during the pre-stimulus (background) period were larger in the young subjects than in

the elderly at the parietal and occipital electrodes. In the elderly subjects there was a

lower rate of deflections during the period of evoked activity (50-350 ms after the

stimulus) compared to that during the 300 ms pre-stimulus (background) period.

There was no such difference in rates of deflections in the young subjects. The overall

period of evoked activity at electrode Oz began significantly earlier in the young

subjects and this seemed to be due to the presence of an earlier positive peak at about

70 ms. That was not seen in the elderly. Analysis of three VEP components (N1, P1

Page 13: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

13

and N2) using the Time distributions of deflections revealed significantly greater

variability in the elderly for components N1 and P1, as well as significantly higher

standard deviations of mean latency for all three components (N1, P1 and N2). The

reverse was true for the young, who demonstrated a greater degree of stimulus time

locking, and narrower time regions related to the VEP components N1, P1 and N2 at

electrode Oz. The young also had significantly higher peak amplitudes for the

deflections within the Amplitude Profiles related to components N1 and P1 than in the

elderly. Amplitudes measured using the Amplitude Profiles were significantly higher

than the amplitudes found in the conventional averages derived from the same data,

illustrating the distortion caused by averaging. In addition the ratios of the mean

amplitudes of the evoked period Amplitude Profiles compared to the mean amplitudes

during background activity (pre-stimulus) for all three response components were

greater than one, indicating that there is amplification during the evoked period. This

degree of amplification (ratio) was larger in the young subjects for N1 and P1 and the

difference was significant for wave N1. Not all stimulus trials contributed deflections

to the evoked activity components and these two subsets of stimulus trials could be

analyzed separately.

Conclusion - The novel methods provide a great deal of additional information

about age-related differences between young and elderly in cortical activity related to

the response and to background activity that is unavailable when analyzing data using

only conventional averaging. This technique has succeeded in differentiating between

background activity and evoked activity, in identifying the onset and the termination

of each component of the evoked activity, indicating which specific stimulus trials

gave rise to a detectable response to the stimulus and which did not. This has lead to

statistical evaluations of the time and amplitude variations of single trial responses.

Finally, the technique has provided insight into the mechanisms of generation of EP,

Page 14: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

14

showing that a major factor is the time locking of the random higher frequency

ongoing EEG deflections, and in addition, their amplification. The conventional EP

technique is also improved by using only those stimulus trials which contribute

deflections to the various response components. Therefore it is likely that it will allow

researchers to access important information concerning underlying physiological

differences in a wide range of groups as well as various types of neuropathology that

would not otherwise be possible using the more conventional forms of analysis.

Abbreviations

APs - action potentials

ECG - electrocardiogram

EEG - electroencephalogram

EMG - electromyogram

EOG - electro-oculogram

EP - evoked potentials

EPSP - excitatory postsynaptic potentials

ERPs - event related potentials

IPSP - inhibitory postsynaptic potential

PREPs - pattern reversal evoked potentials

SNR - signal to noise ratio

SL - subjective threshold

VEP - visual evoked potentials

Page 15: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

15

Technique for the Analysis of Evoked and Background EEG Activity applied to

the Young and Elderly Subjects

Introduction

A technique used for the exploration of brain functioning in humans is the

recording of electrical activity from the surface of the scalp with the use of electrodes,

known as the electroencephalogram (EEG), which represents brain activity

accompanied by electrical changes and if there is a sensory stimulus by evoked

potentials (EP). Electroencephalography provides a noninvasive method for the study

of ongoing or spontaneous electrical activity of the brain. The origin of the changes in

electrical activity found in the EEG is linked to the structure and function of the cells

in the cerebral cortex which are generating the activity. Neurons are the primary

source generators of electrical activity in the cerebral cortex that is found in EEG and

EP. It is estimated that the cerebral cortex contains a total of approximately 5x1010

neurons, where a slab of cortex with a volume of 1mm3 contains approximately

40,000 neurons, and 88-108 synapses. Each neuron receives approximately 20,000 pre-

synaptic inputs, 18,000 of which are excitatory and 2,000 are inhibitory (Kiloh et al.,

1981).

The EEG reflects changes of a steady electrical charge on the cell membrane of

cortical neurons, as a result of impulses arriving from other neurons via axons which

terminate in a synapse, which in turn release a neurotransmitter across the synaptic

cleft to the postsynaptic neuronal membrane that produces changes in the membrane

potential. The interaction of the neurotransmitters with postsynaptic receptor sites

produce a transient change in the membrane potential. The membrane potential either

depolarizes, which is known as an excitatory postsynaptic potentials (EPSP) that

occurs due to an increased permeability of the membrane to sodium and other ions at

Page 16: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

16

the receptor zone, or hyperpolarize the membrane potential, known as an inhibitory

postsynaptic potential (IPSP) that is caused by an increased permeability to potassium

and chloride ions, both of which occur in cortical neurons as a result of thalamo-

cortical and other input. Much of the rhythmic activity seen in the on-going EEG in

various physiological states in influenced by a thalamic pacemaker which during

“activation” or a change in the physiological state of the subject may abolish the

rhythmic discharges in the thalamic nuclei causing cortical potentials to become

desynchronized (Kiloh et al., 1981). These postsynaptic potentials can last up to over

100 ms, and are subsequently summated on different parts of the cell body (for review

see Hillyard and Picton, 1987). The ongoing EEG that is recorded at the scalp

represents these synaptic dendritic potentials. Much of the postsynaptic activity is

caused or “triggered” by action potentials (APs). Apparently APs do not contribute to

the extracellular potentials since the “spike” that is produced is brief (approximately

1-3 ms) and rarely demonstrates synchronous firing amongst adjacent cells, thus being

unlikely that any sufficient summation of the potentials occur that can be detected at

the surface of the scalp (Niedermeyer and Lopes da Silva, 1987).

The electrical activity can be recorded through the use of scalp electrodes since

electrical fields are primarily generated by the cerebral cortex, which is the outer layer

of the cerebrum, where the neuronal dendrites extend through most layers of the

cortex, and the pyramidal cell neurons tend to be closely packed with apical dendrites

are oriented parallel to each other, facilitating spatial summation of currents generated

by groups of neurons, creating electrical fields, which are referred to as field

potentials. Some patterns of cell membrane activation produce only local currents or

“closed” electrical fields, and therefore are unable to generate potential fields at a

distance. In contrast, an “open” field is generated when currents flow beyond the

Page 17: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

17

limits of groups of active cells and their processes. This allows part of the current to

pass through the meningeal coverings, spinal fluid, and skull to the scalp where it can

be recorded (Hillyard and Picton, 1987).

The EEG (and EP) is usually recorded by applying electrodes to the scalp. A

‘montage’ is used to connect the electrodes to different recording channels. Typically

a ‘referential’ montage is used where each channel records the difference between one

scalp electrode and a common reference. Electrical fields are continuously being

generated within the brain, emitting potentials that are spread to all areas of the scalp.

Therefore, there is no location on the scalp at which some electrical activity cannot be

recorded. This requires researchers to find an ‘inactive’ reference electrode location

that does not reflect electrical activity from the cortex. Researchers tend to use the

linked-ear (mastoids), while others use the nose tip as reference point electrodes. Also

bipolar montages are used, which record the electrical activity between two electrodes

(Picton et al., 1995). The international “10-20 system” is typically used for the

placement of electrodes for neurophysiological recordings at fixed, internationally

agreed upon sites, that were proposed by the American Electroencephalographic

Society (1991) (See Figure 1).

Page 18: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

18

Figure 1. Diagram of the international “10-20 system” placement of scalp electrodes for neurophysiological recordings at fixed, internationally agreed upon sites, that were proposed by the American Electroencephalographic Society (1991). Electrodes consisted of sixteen mono-polar electrode sites: Fz, F3, F4, Cz, C3, C4, T3, T4, T5, T6, Pz, P3, P4, O1, O2, and Oz, with the reference electrode placement being the left mastoid. The ground electrode was located on the forehead. Electrodes were placed around the eyes in order to record the vertical and horizontal electro-oculogram (EOG).

The amplitude of the electrical signals that are recorded via scalp electrodes are

small (tens to hundreds of microvolts) and need to be amplified. Differential

amplifiers are typically used to amplify the electrical signals picked up by the scalp

electrodes. The electrical signals are connected to the input stage via two leads.

Because the signal is amplified between two input leads, it is referred to as being in

differential mode. A balance between the two halves of a differential amplifier is

Page 19: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

19

maintained throughout the recording which is intended to minimize the amplification

of any common mode signals such as mains (line) interference, which is in phase at

both input leads, and thus tends to be cancelled out. Filters are also used to exclude

recorded electrical activity of relatively high or low frequencies below and beyond the

range of EEG frequencies (0.5-40 Hz). Filters can aid in providing a clearer

representation of the EEG activity with minimal distortion from electrical sources not

related to cortical activity (Hillyard and Picton, 1987).

A source of noise that is prevalent in EEG and EP recordings is the ocular

artifact. Ocular artifacts represent electrical activity due to eye movements and blinks

that occur simultaneously with the EEG or EP waveform. A technique that is aimed at

minimizing ocular artifacts is the recording of the electro-oculogram (EOG). The

EOG is recorded using separate electrodes that are fixed symmetrically above and

below and on either side of the eyes in order to directly monitor eye blinks and eye

movements. The EOG recording enables the use of linear regression to correct the

EEG waveform for the presence of ocular artifacts (Semlitsch et al., 1986). Other

gross contaminations of the EEG recordings produced by tongue, and jaw movements

can also occur. These artifacts must be controlled when conducting EEG recordings

by instructing subjects not to swallow, or make any head movements, and to keep

perfectly still during the recording session. ECG potentials (electrocardiogram) may

be minimized by ensuring that the reference electrode placement is optimal (i.e. not

placed on the neck or surrounding areas), as well as changing the subjects posture

during recordings.

Normal EEG consists of waves of various frequencies and amplitudes that

depend on the subject’s physiological states, such as wakefulness, the age and health

of the subject, as well as the location of the recording electrodes. The EEG activity

appears as periodic waves with frequencies ranging from 0.5 to 40 cycles per second

Page 20: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

20

(Hz), and with amplitudes that range from five to several hundred microvolts. EEG

activity has been categorized into four frequency bands: delta (0.5 to 3.5 Hz), theta (4

to 7 Hz), Alpha (8 to 13 Hz), and beta (14 to 30 Hz). Alpha rhythm amplitudes tend to

be greater (up to 50 V) in relation to beta rhythm amplitudes which are usually very

low (5 to 30 V). As a rule, as the frequency of the EEG activity increases the

amplitudes decrease. Recordings of a normal adult subject in a relaxed state with eyes

closed will be characterized most dominantly by alpha activity, but if the subject is

instructed to open his or her eyes the activity will become less synchronized and

increased in frequency producing dominant beta activity that is also characterized by

lower amplitudes. Delta and theta wave activity are most prominently found during

sleep (Willis, 1998) (see Figure 2 - for typical EEG with different frequencies).

Figure 2. Example of typical EEG in the four frequency bands: delta (0.5 to 3.5 Hz), theta (4 to 7 Hz), Alpha (8 to 13 Hz), and beta (14 to 30 Hz).

Page 21: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

21

Evoked Potentials

The technique of recording evoked potentials (EPs) using EEG from an intact

scalp is one of the few available investigative techniques that provide precise

information on the time course of the neural activity that underlies higher brain

functions. The EPs reflect voltage changes in the ongoing EEG that are time locked to

and initiated by sensory, motor, or cognitive events. Much like EEG and APs, EPs are

primarily generated when there is a depolarization and possibly also when there is a

synchronous hyperpolarization of cell membranes in the nervous system. The

electrical activity recorded by the scalp electrodes includes not only the EP (signal),

but also other ongoing, spontaneous, random electrical activity (noise) such as that of

the heart (ECG), of the muscles (electromyogram-EMG) and the ongoing EEG itself.

The amplitude of the EP signal is usually smaller than that of the noise. Therefore

there is a small signal to noise ratio (SNR). In order to improve this SNR, it is

necessary to optimize the magnitude of the signal while at the same time reducing the

noise. In order to augment the signal, stimuli which synchronously activate the EP

generators are used. Temporal synchrony is an important factor in generating field

potentials (leading to EP) that are produced by a group of cells. When there is a

synchronous activation of geometrically organized cells related to a stimulus with a

sharp onset, a summation of the fields generated by individual cellular elements

occurs, that is similar in sign and orientation. This summation increases the amplitude

of the field potentials. Examples of sensory stimuli which lead to synchronous

activation of neuronal elements include clicks (auditory modality) and flashes or

alternating checkerboards on a screen (visual). In order to differentiate between EPs

and EEG ongoing activity, it is necessary to use a filter that will pass the frequency

range of interest (that of the EP) and not the range above or below that range, as well

Page 22: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

22

as averaging many trials. With respect to the averaging procedure, it is assumed that

the EEG activity is random thus being different for each repetition of the event (or

trial) and subsequently decreases in amplitude with averaging. On the other hand, the

EP is time-locked to the stimulus (appears after a more or less fixed latency) so that

the EP is summated over repeated stimulus presentations and averaging. Although the

background noise is never completely eliminated from the recorded activity, the SNR

is greatly improved.

In general, one can divide the EPs recorded from the scalps of human subjects

into two classes of EPs. One is exogenous in nature meaning that electrophysiological

activity is determined by the character of the stimulus (e.g. its modality, intensity, and

repetition rate). Reports in the literature suggest that exogenous EP components are so

consistent across subjects that small deviations from the typical structure are very

reliable in indicating any sensory or neurological dysfunction (for review see Hillyard

and Picton, 1987). The other form of cortical activity is endogenous in nature,

meaning that the activity is related to the cognitive reaction, or attitude of the subject

in response to a stimulus. These endogenous waves are also called event related

potentials (ERPs) and are believed to reflect more cognitive processing in humans (for

review see Picton et al., 1995). Exogenous EP components tend to have shorter

latencies than the endogenous ERP components. Our discussion will be limited to

exogenous EP components due to the nature of the stimulus used in the paradigm of

this study. The exogenous EP can be further broken down into twp types: action

potentials (AP) and synaptic-dendritic potentials. For example, the auditory nerve-

brainstem evoked potentials and the initial periods of the somato sensory EP represent

the compound AP of axons synchronously activated. Therefore the recording filter

passes the higher frequencies of APs (e.g. 100-3000 Hz); requires faster A to D

Page 23: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

23

conversion, a shorter time window, and short time periods per bin (e.g. up to 100 s).

Thus the EEG is filtered out. The synaptic dendritic EP represent lower frequency

electrical activity so that the recording filter passes lower frequencies (e.g. up to 100

Hz), a longer “window” (e.g. 1000 ms) and a longer time period per bin. Much of this

type of EP is generated in the cerebral cortex and the cortical visual evoked potentials

(VEP) will be the type of EP analyzed here.

Nomenclature

There are several systems of nomenclature that exist which attempt to classify

the many complex components contained within an EP waveform. Thus, the principal

waveform components are characterized by a specific polarity (positive or negative),

an amplitude that is measured in V, and a latency (time elapsed after the onset of

stimulus). Therefore, the EP is viewed as a sequence of serially activated processes

that are manifested as distinct positive-negative potential fluctuations. For example, a

P100 component typically found within particular VEP waveforms is a positive wave

occurring at approximately 100 ms after the stimulus.

Because the latency of an EP peak may vary from one subject to another, the

average latency of the peak across the normal subject population is often used to

designate the component. Under different task conditions and amongst different

groups of subjects, there is a wider variability in the latency of the endogenous than of

exogenous components (for review see Hillyard and Picton, 1987; Picton et al., 1995).

Analysis of EEG and Evoked Potentials

The two most common approaches to the quantitative analysis of EEG and EP

activity utilize the methods of the averaging of responses to repeated stimuli (for EP),

and frequency analysis (for EEG and EP). These methods have certain limitations,

which reduce their utility in the analysis of single evoked potentials. The conventional

averaging technique for EP evaluates EEG segments after repetitive stimulus trials

Page 24: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

24

assembled in a common store in order to reveal the components invariant in each trial.

This technique improves the SNR, thus differentiating the time-locked stimulus

specific response from the random on-going EEG background activity. This is

accomplished however at the cost of losing valuable information, such as moment-to-

moment variation with respect to time (i.e.: jitter), and similarly with respect to

amplitude. It is assumed that these variations contain important information about

brain function, and it would be helpful not to ignore them, if possible, when

investigating different groups or pathologies. Frequency analysis can be problematic

in regards to cortical brain activity over a period of time, since the integrative aspect

of this form of analysis does not address moment-to-moment variations.

These problems can be overcome if a way could be found to conduct single

response trial analysis. Several procedures have been suggested by previous authors

for the analysis of single trials. The classical approach (Woody, 1967) suggests that

due to the similarity between the waveforms of the averaged EP and single responses,

the averaged EP can be used as a criterion template for the selection of the appropriate

trials for the single trial analysis. Thus only the components of EP which are revealed

after averaging, can be used. However, it is known that some components can be

distorted, highly decreased in amplitude or even lost after averaging, leaving a deficit

of possibly important information needed for the study of single trials. Ford and

Pfefferbaum (1991) and Ford et al. (1994) elaborated on the procedure proposed by

Woody (1967) with a signal/noise screening procedure by matching the EEG template

not only to the time period where the signal is expected but also to a later range

(noise) where no signal is expected. Trials that did not pass the screen were rejected,

thus possibly excluding valuable information. Despite this and other work that has

been done in this area, the problem of single trial analysis related to background EEG

Page 25: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

25

activity has yet to be solved. In addition these methods often involve complex and

unwieldy mathematical computations and extensive analysis.

Recent studies have demonstrated a connection between background EEG

activity and EPs. These results lead to the suggestion that the processing and analysis

of EEG and EP should be conducted with the same method for both aspects of the

data (Jansen and Brandt, 1991; Rahn and Basar, 1993; Arieli et al., 1996; Polich,

1997; Basar et al. 1997; Basar et al., 1998; Kisley and Gerstein, 1999). Jansen and

Brandt (1991) examined the effects of pre-stimulus alpha activity on the averaged

visual evoked response to low intensity flashes. There was a relationship between the

averaged visual evoked potential latencies and amplitudes of components N1 and P2

with the phase of the alpha activity immediately preceding the stimulus, where N1

appeared to be entrained alpha activity, while the P2 component was not an alpha

process, but was influenced by the amount of pre-stimulus alpha activity. Rahn and

Basar (1993) compared conventionally averaged VEPs in response to flash stimuli

with the alpha and theta band EEG activities preceding the stimulus presentation.

They found that the VEP amplitudes were affected qualitatively depending on the

frequency content of the pre-stimulus activity. Arieli et al. (1996) recorded real-time

optical imaging, as well as local field potentials and spike discharges of single

neurons in the cortex of the adult cat to repeated presentations of the same stimulus,

analyzing spatio-temporal dynamics in single trial responses to visual stimulation.

This was done using moving gratings, and recorded from microelectrodes inserted

into an exposed area of the visual cortex. It was suggested that the variability or

changes in the patterns of the evoked activity from trial-to-trial are caused by

fluctuations in the ongoing cortical activity, which appears to have a major influence

on sensory processing. In a related study, Kisley and Gerstein (1999) conducted

Page 26: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

26

electrophysiological recordings of local field potentials and unit spike trains in the

auditory cortex of ketamine/xylazine-anesthetized rats, using principal components

analysis to quantitatively classify waveforms on the basis of their time courses. Trial-

to-trial variability seemed to be dependent on ongoing background activity, which

appeared to modulate both the amplitude and latency of the evoked local field

potentials and evoked unit activity. Similarly, Polich (1997) reported that variation of

the background EEG contributed to the individual variability of the ERPs (P300

component) using an auditory oddball paradigm, recorded with both eyes open and

eyes closed. The relationship between these two domains was attributed to an

association between the spectral power and mean frequency of the EEG background

activity and amplitude and latency of the P300 component in adult humans. Basar et

al. (1997) and Basar et al. (1998) who studied spontaneous EEG activity in relation to

auditory and visual evoked potentials as well as cross-modality experiments in young

human subjects, suggested that spontaneous EEG alpha (10-Hz) activity is not pure

noise as was previously considered in the past, but rather the evoked alpha oscillations

patterns which are initiated by but not time-locked to a stimulus.

New Methods of Analysis

This study employs new procedures for the analysis of single trials that is not

based on the conventional averaged EP. The principal tool of this approach is the

statistical description of the appearance of deflections (peaks) in the ongoing EEG

before and after the onset of sensory stimuli. An attempt was undertaken to build

representative distributions of the on-going background and evoked activity similar to

the post-stimulus histogram used in neurophysiology to describe a sequence of spikes.

Therefore, the data required for the new analysis techniques involved the detection of

all positive and negative deflections in the ongoing EEG activity before and after the

Page 27: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

27

onset of sensory stimuli. This method enables the researcher to examine the time

distribution of the EEG events (peaks, waves) only, without any relation at this stage

of the analysis to their amplitude. Such time patterns (a time distribution of the

appearance of positive and negative deflections) have been termed Time Distributions

of Deflections and are constructed for positive and negative deflections separately.

These distributions are used for the analysis of the two categories of scalp recorded

activity, one of which is defined as originating from the response to specific stimuli

(evoked activity), and the other being defined as background activity which would

appear before the stimulus, thus making use of the same methods of analysis for both

background and evoked activity, which was problematic for previously proposed

methods. The distributions provide a comprehensive representation of all the activity

found within a particular number of single trials, without having to incorporate

various estimation techniques that are characterized by inherent limitations and

weaknesses from lack of knowledge of the SNR. Thus, these new types of analysis

contribute to the differentiation between deflections related to responses evoked by

the stimulus and deflections representing the background activity, and the comparison

between them, possibly providing an initial approach towards solving the long

standing problem of single trial analysis. This offers an alternative to conventional

averaging while maintaining information as to the overall variability of this cortical

activity. In addition to the analysis of Time Distributions of Deflections, the mean

amplitudes of the detected deflections within each successive time bin can also be

analyzed. This second analysis should provide information about the magnitude of the

activity evoked by the stimulus and of the background activity.

To generate the evoked activity that will be analyzed using the new methods of

analysis, a visual stimulus was used in this paradigm in the form of pattern-reversal

Page 28: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

28

checkerboards. The principal exogenous waves of visual EPs that will be analyzed are

components N1, P1, and N2. Thus we will be examining not only background

activity, but also its relation to visual evoked potentials (VEP). Thus the new

techniques for analysis of the EEG and EP take into account the exact polarities, times

and amplitudes of each deflection of each stimulus. Therefore the techniques bring us

closer to a single trial analysis and provide information not usually accessible when

using standard averaged EP techniques.

Age-Related Background

After elaboration of these new techniques as part of this study, we hope to

explore the utility of these new types of analyses to the analysis of EEG and EP in

young and elderly subjects, using the pattern-reversal checkerboard stimulus. The

study of age-related neurophysiological changes is greatly facilitated by using EEG

research, which provides a better measure of temporal aspects of information

processing than do behavioral measures (Taylor, 1995). Applications of the newly

proposed technique may be explored by comparing the young and elderly and

differences can be expected to occur based on numerous biopathological changes

which are associated with normal aging. The basis for expecting differences in EEG

between the young and elderly is largely based on findings of results obtained from

the EEG by spectral analysis. Age related differences of EPs are largely based on

studies that compared conventional averaged EPs in two groups of subjects (young

and elderly). We would expect that this new technique for the analysis of EEG

activity would reveal important differences in both evoked and background activity

between the young and the elderly subjects.

Evidence from autopsies and computerized tomography scans show that the

weight and volume of the brain diminishes continuously with age even in healthy

Page 29: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

29

individuals (For Reviews see Braak and Braak, 1988; Giaquinto, 1988). The aging

process has been referred to as a continuous and uniform slowing of neural

transmission (synaptic delay) which is characterized by changes in neurotransmitter

function such as decreases in the amount of both excitatory and inhibitory

transmitters, and/or their associated enzymes (for review see Iwangoff et al., 1980;

Giaquinto, 1988; Giacobini, 1990; and Schroder et al. 1991). It has also been

suggested that these various age-related biopathological changes mentioned above

cause a decrease in the conduction velocity in the peripheral nerves and central

pathways which has been widely reported with regards to the elderly (Downie and

Newell, 1961; Buchthal and Rosenfalck, 1966; Celesia and Daly, 1977, Drechsler,

1978; Dorfman and Bosley, 1979, Kazis et al. 1983). In addition, results obtained

from the EEG by spectral analysis show changes in various frequency ranges in old

age. [For a comprehensive review see Dustman et al., 1993.] Age-related dendritic

changes of the isocortical pyramidal neurons in the elderly which result in the

impairment of dendritic processes include: swellings of the pyramidal cell soma,

progressive loss of dendritic spines as well as irregular nodulation of the dendrites.

Related changes affect the visual pathway system as well (For Review see Dolman et

al., 1980, Braak and Braak, 1988). It has also been suggested that the visual system in

the elderly is also affected by loss of retinal ganglion cells (axonal dystrophy) in the

optic nerve and optic pathways (Vrabec, 1965; Dolman et al., 1980; Gartner and

Henkind, 1981), as well as other age-associated changes occurring in the optic nerve

(Dolman et al., 1980). Considering the many age related physiological changes, the

study of their underlying processes using this new method of analysis may yield

important information as to the electrophysiological expression of these changes.

Page 30: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

30

Visual evoked-potential (exogenous components)

The studies that examined visual evoked potentials (VEPs) that are directly

relevant to the work presented here are paradigms that employed the widely used

pattern reversal checkerboard, which elicits what is known as pattern reversal evoked

potentials (PREPs). The alternating checkerboard pattern is a strong stimulus to the

visual system (and therefore elicits high amplitude responses synchronously – VEPs)

since, as shown by Hubel and Wiesel (1959, 1968, 1974), the visual system is

extremely sensitive to high contrast stimuli and the alternating checkerboard provides

high contrasts. Pattern reversal stimuli responses are elicited by presenting subjects

with a pattern of checkerboard squares to look at, in which all black squares suddenly

turn white and all white squares suddenly turn black. Each pattern reversal causes a

response in the form of a positive peak at approximately 100 ms (P100), as well as

other components (i.e. negative components N1 and N2). The response is dependent

also on the retinal area that is stimulated, and luminance, as well as the individual

subjects visual acuity, attentiveness, and recording electrode location. If these

variables are kept constant, the latencies of the responses to the pattern reversal

stimuli tend to show little variation across normal subjects of the same age. The

choice however to use pattern reversal stimuli versus flash visual stimuli was based on

the fact that flash responses vary widely between subjects (the major positive

component comparable to the pattern reversal stimuli P100 or P1 component tends to

vary in latency between 50-100 ms), and is not very effective in detecting

abnormalities, which is not the case with respect to the pattern reversal stimuli. VEPs

or PREPs are believed to originate from the extrastriate cortex (Halliday and

Michael, 1970; Jeffreys and Axford, 1972; Lesevre and Joseph, 1979), as well as

being mediated by extrageniculo-calcarine pathways to secondary visual cortices,

namely extrastriate areas 18 and 19 (Celesia et al., 1980). A negative wave that

Page 31: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

31

precedes the P1 wave at approximately 70-90 ms is typically known as N1, which is

believed to reflect activity generated within the primary visual cortex (striate cortex -

area 17) (Cobb and Dawson, 1960; Michael and Halliday, 1971; Lesevre and Joseph,

1979; Celesia et al., 1980; Jefferys and Axford, 1972). The next major exogenous

component of interest that follows the P1 is a negative component known as the N2

that typically ranges in latency between 130-190 ms and is believed to reflect activity

within the extrastriate cortex (Michael and Halliday, 1971; Lesevre and Joseph, 1979;

Celesia et al., 1980) much like the activity of the P1 component, but where P1 activity

is believed to reflect activity in the visual association area 19, the N2 is believed to be

more related to area 18 (Jefferys and Axford, 1972) (see Figure 3 for example of VEP

waveform).

Page 32: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

32

Figure 3. Example of a typical conventionally averaged visual evoked potential waveform from our data which consists of two negative components: N1 and N2 and one positive

component: P1. X-axis: time after stimulus. Y-axis: V.

Page 33: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

33

Goals

The main purpose of this investigation is to describe the technique and to

present the first results using the new techniques of analysis of exogenous

components elicited by a visual stimulus for young and elderly subjects, which allows

for the differentiation and evaluation of EEG background activity and evoked activity,

and to examine the mechanisms that contribute to the generation of visual EPs.

Importance

The new techniques provide a clearer understanding of underlying age-related

physiological changes as well as information related to the generation of evoked

activity, and may provide insight into cortical changes that are associated with various

forms of neuropathology.

Methods

Subjects

Participants in the study were 14 young subjects ages 18-35, with average age

25.4 (S.D. ±4.0), mean education 14.2 years (S.D. ±1.1) and 14 elderly subjects ages

65-80, average age of 75.2 (S.D. ±6.0), and a mean education of 14.9 years (S.D.

±2.8). All participants volunteered in response to advertisements in local newspapers,

or through referral. Participants were screened by responding to health questionnaires

in order to identify any physical or medical limitations that would disqualify them for

participation in the study. All subjects reported being in good health with 20/20

(corrected) vision and normal hearing, with no history of neurological or mental

illness. The experimental protocol was reviewed and approved by the committee for

human experimentation of the Hebrew University- Hadassah Medical School.

Stimuli

Auditory and visual stimuli were presented in alternating order, each at a rate of

Page 34: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

34

0.5/sec. The duration of a stimulation/recording session was 10 minutes (300 auditory

and 300 visual stimuli: Aud-Vis-Aud-Vis…). Six sessions on two experimental days

(three sessions per day) were conducted with each subject. To ensure subject

attentiveness, each was instructed to perform mental counting of each perceived

auditory stimulus. The auditory stimulus was a binaural click of 30 dB above his

subjective threshold (i.e. SL). The visual stimulus (see Figure 4) consisted of an

alternating checkerboard with a full field of 11 o (25.7 cm.) and an angle for check

viewing of 1.35 o (81’ check size) (note the rather low basic spatial frequency of the

visual stimuli). The actual activating visual stimulus was the pattern reversal (40 s for

the reversal to occur) and not the steady state pattern present on the screen for 2 seconds.

Subjects were sitting 140 cm. from a 27x21 cm. CTX computer screen, which provided

the visual stimuli. In three of these subjects a session was conducted during which

sensory stimuli were not presented. This report describes the analysis of the responses to

the visual stimuli only. The rationale behind using and auditory and visual bimodal

stimulation in our experimental design was motivated by its efficiency with respect to

number of recordings, thus reducing possible added fatigue factors that could arise

from conducting separate (thus longer) unimodal stimulus recordings. There was no

specific rationale behind the choice to analyze the visual stimulus versus the auditory,

but only one stimulus modality could be included in the analysis of this study due to

limited resources. It is possible to conduct an analysis of the auditory responses in the

future.

Page 35: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

35

Figure 4. Diagram illustrating the sequence of stimulus presentations. Stimuli were visual and auditory clicks presented in alternating order (visual- auditory-visual---). The visual stimuli were pattern reversal checkerboards with full field of 11º (25.7 cm). The checkerboard was on throughout and reversed every two seconds. The auditory clicks were delivered at the midpoint in time of each pattern presentation at 30 dB above subjective threshold, presented by earphones at a rate of 1 every 2 seconds.

Apparatus and Procedure

Scalp activity was recorded with an Electrocap (Neuroscan, Inc., Sterling, VA,

U.S.A.) using sixteen monopolar electrode sites: Fz, F3, F4, Cz, C3, C4, T3, T4, T5,

T6, Pz, P3, P4, O1, O2, and Oz of the international 10-20 electrode placement system

(see Figure 1) with the reference electrode at the left mastoid, and the ground

electrode was located on the forehead. Vertical and horizontal electrooculograms

(EOG) were recorded in order to control for eye movement artifacts. EEG was

corrected for the presence of vertical EOG artifact by linear regression (Semlitsch et

al., 1986). The stimulus presentations and data acquisition were provided by a

STIM/SCAN EEG/EP workstation (Neuroscan, Inc., Sterling, VA, U.S.A.). EEG and

EOG channels were continuously recorded through Synamps Amplifiers (bandpass =

0.1 to 100 Hz) and digitized at a rate of 1000 Hz. Data were stored and analyzed off

line. Although there are many different methods in which electrodes can be attached

Page 36: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

36

to the scalp, the electrode cap is the most widely used method in EEG and EP

research. The cap is made of an elastic material in which plastic mounts are attached

containing pure tin electrodes that have been strategically placed in such a way that

when the correct size cap is properly fitted, the electrodes are on the 10-20 position

sites.

Experimental confounds were controlled for by maintaining strict procedures

and laboratory conditions during the recordings, such as maintaining the same level of

lighting at the time of the recordings.

Data Processing

The 1,300 ms EEG segment recorded beginning 300 ms before the visual

stimulus and continuing 1000 ms after stimulus presentation was defined as a single

trial. The frequency content of the recorded activity was limited by an additional

bandpass filter (1-30Hz, 24 dB/octave). In addition, smoothing of traces was

conducted across every 5 data points. Computer algorithms were used to identify all

successive positive and negative deflections, no matter how small in amplitude,

ignoring the lower frequency fluctuations of the EEG upon which the higher

frequency deflections may be superimposed. The time of their peaks during the 300

ms pre-stimulus and the 1,000 ms post-stimulus periods was determined (see Figure

5A). The base to peak amplitude of each such deflection was also measured. The time

coordinates of the deflections with respect to the onset of the stimulus trigger were

determined separately for positive and negative deflections for each trial (see Figure

5B).

Page 37: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

37

Figure 5. Scheme of the description of the EEG deflections. A: Times of the EEG peaks during the 300 ms pre-stimulus and the 1,000 ms post-stimulus period, showing the time distribution of deflections. B: Base to peak amplitude of each deflection, separately for negative and positive deflections.

Page 38: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

38

Thus each single trial was described by a sequence of events (deflections), each of

which had a time coordinate (with respect to the stimulus), polarity and amplitude.

The number of positive and negative deflections was separately calculated in bins of 6

ms for a series of up to 300 trials, and was defined as the Time Distribution of

Deflections for the given subject, given number of trials and given session (see Figure

6).

Page 39: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

39

Figure 6. The Time Distribution of Deflections. Subject SB (1 session), electrode Oz.. The Time Distributions of Deflections - the number of positive and of negative deflections obtained in bins of 6 ms each for a series of 30, 100 and 300 trials. The lowest panels are the mean time distributions from all 6 sessions (1,800 trials) and the conventionally averaged VEP derived from the same data. X-axis: the time coordinate of the time bins with respect to the onset of stimuli - time 0 = trigger of the visual stimulus. Y-axis: the number of deflections obtained per time bin, normalized by the number of trials used (percentage).

Page 40: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

40

Because of the filtering (1-30 Hz) and subsequent smoothing (smoothing of traces

was conducted across every 5 data points), there can be no more than one deflection

in a single 6 ms bin during a single stimulus trial. Thus one trial could contribute no

more than one deflection per time bin (6 ms). The Time Distribution of the positive

and negative (separately) deflections obtained over the six different recording sessions

from the same subject could be combined (averaged or summated, wherever

appropriate) into a composite Time Distribution of Deflections. There are thus for

each subject 1,800 stimulus trials (6 sessions, 300 stimulus trials in a session)

contributing to this composite time distribution of deflections. The ordinate (y axis) of

this time distribution is sometimes shown as the total number of deflections per bin

and occasionally as the percentage of trials in that session with deflections in each bin.

In addition, the base to peak amplitudes of these same positive and negative

(separately) deflections within each time bin as a function of pre- and post-stimulus

time were determined and summed, and finally divided by the number of deflections

detected in each such time bin. Note that the number of these deflections may differ

significantly from bin to bin, especially during the period of evoked activity. This

gave the average amplitude per time bin (for each polarity). When this is done for the

300 stimulus trials in a session, for 6 sessions in a given subject, an Amplitude Profile

is obtained (see Figure 7).

Page 41: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

41

Figure 7. The Amplitude Profiles. Subject SB (1 session), electrode Oz. The Amplitude Profiles.- the mean amplitudes of the positive and of the negative deflections within each time bin as a function of pre- and post-stimulus time for a series of 30, 100 and 300 trials. The lowest panels show the mean amplitude profiles from all 6 sessions (1,800 trials) and the

conventionally averaged VEP from the same data. X-axis - the time coordinate of the time

bins with respect to the onset of stimuli - time 0 = trigger of the visual stimulus. Y-axis – the sum of the base to peak amplitudes of deflections obtained in the given time bin over the set of trials, divided by the number of deflections detected in this time bin, i.e. mean amplitude of

deflections [V] per time bin.

Page 42: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

42

The mean rate of deflections in a given time period (e.g. 300 ms pre-stimulus

period) for a given number of trials was calculated by taking the overall number of

deflections of the same polarity detected in the given time interval in all given trials,

and dividing this by the duration of the analyzed period and the number of trials.

These time distributions of deflections and amplitude profiles provided the

basis for further analysis.

The same recorded EEG, after filtering 1-30 Hz, was also subjected to

conventional averaging, and the time distributions of deflections and the Amplitude

Profiles obtained with this new technique were compared to the conventionally

averaged EPs obtained from the same subjects, from the same recording sites over the

same time period and for the same components. The latency of the conventionally

averaged VEP N1 component was defined as the most negative point between 55 and

95 ms after the stimulus, P1 as the most positive point between 75 and 135 ms, and

N2 as the most negative point between 130 and 206 ms.

The EEG recorded in three of these subjects in the sessions without sensory

stimuli was analyzed and compared in the same subjects to the pre-stimulus period in

the sessions with the sensory stimuli using the new technique in order to specify the

time and amplitude distributions of “true” background activity. This would help in

differentiating the time period which is truly a “response” in nature.

These new analysis techniques (time distribution of deflections and amplitude

profiles) can be used to define background activity and evoked activity, to derive

aspects of the time variability of response activity, amplitude variability, the

differentiation between trials containing a detectable response to the stimulus (single

trial analysis) and those which do not. This technique is protected by a provisional

patent from the U.S. Patent and Trademark Office (60/316,974; Sept. 5, 2001).

Page 43: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

43

Results

Differentiation between background and evoked activity - Time distributions and

amplitudes of the EEG deflections: Criteria for determination of the evoked response

applied to young and elderly subjects

This new analysis technique is based on a statistical description of the times of

background deflections (the time distribution of deflections) and of the amplitudes of

these same deflections (the amplitude profile). Such distributions and profiles, along

with the conventionally averaged VEP derived from the same data for a typical young

and a typical elderly subject are shown in Figure 8. One can see clear parallels

between the typical conventionally averaged VEP components (N1, P1, N2) and the

corresponding waves in the time distributions and amplitude profiles in the subjects.

Page 44: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

44

Figure 8. Time Distribution of Positive and Negative Deflections, Positive and Negative Amplitude Profiles, and conventionally averaged VEP from a typical young (left) and a typical elderly (right) subject. X-axis: Time before and after the onset of the stimulus [ms]- time 0 = trigger of visual stimulus reversal. Y-axis: Time distributions: Percentage of deflections per time bin, normalized by the number of trials used. Positive and Negative Amplitude Profiles:

Y-axis: Mean amplitude of deflections per time bin in V. VEP waveforms: Y-axis: amplitude

in V.

Page 45: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

45

It is important to note that even though more than one recording session was

conducted in this study in order to assess reliability and consistency of the data, in

final analysis there was no need for such lengthy recordings. The consistency across

recording sessions was determined by visual inspection of the Time Distribution of

Deflections and the Amplitude Profiles of all six separate recordings superimposed for

each individual young and elderly subject (see Figure 9). It is clear that the traces

recorded in 6 separate sessions on three different days are consistent with each other.

The data in Figures 6 and 7 also suggest that a minimum of 100 trials from one

recording session would be sufficient to receive representative time distributions of

deflections and a minimum of 300 trials from 1 recording session for representative

amplitude profiles of the background and evoked activity for each individual subject.

Page 46: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

46

Figure 9. Six separate recordings of Time Distribution of Positive and Negative Deflections, and Positive and Negative Amplitude Profiles superimposed from a typical young (left) and a typical elderly (right) subject. X-axis: Time before and after the onset of the stimulus [ms]- time 0 = trigger of visual stimulus reversal. Y-axis: Time distributions: Number of deflections per time bin. Positive and Negative Amplitude Profiles: Y-axis: Mean amplitude of deflections

per time bin in V.

Page 47: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

47

One of the primary functions of these novel techniques is the use of the

time distributions of deflections for the differentiation between background and

evoked activity. The time distribution of deflections provides the number of

deflections at any given time period for all of the stimulus trials over the entire

recording session. The first question that must be addressed is how to differentiate

between the EEG deflections that are associated with background activity and

those associated with the response to a stimulus. The 300 ms pre-stimulus period

can be considered as being representative of background activity, since as can be

seen in Figure 8 for electrode Oz (above the primary visual cortex), the deflections

have a “uniform” pattern, i.e. there is no significant difference between the

number of positive and the number of negative deflections across trials at any

given time during this period. On the other hand one may assume that the response

to a stimulus would be characterized by changes in the number of deflections at

specific times and for different polarities (i.e. when there is a disproportionate

increase in the number of positive deflections over the number of negative

deflections or vice versa as can be seen in Figure 8). For example, note that

shortly after 100 ms after the stimulus, the number of positive deflections is much

greater than the number of negative deflections and this takes place at the same

time as the peak of the P1 component of the conventionally averaged VEP derived

from the same data, shown below.

In order to define a statistical criteria for the time of appearance of an evoked

response, one must first define the variance of the background activity for the time

period described above. An evoked response will then be indicated by a significant

increase in this variance above some critical level.

One may assume that the difference between the number of positive and

Page 48: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

48

negative deflections in the time bins representing the background activity across trials

would be small and differ only by chance from zero, i.e. the probabilities of detecting

positive or negative deflections in the particular time bins (across trials) would be

equal. In order to obtain a statistical measure of this variance between the number of

positive and negative deflections and a measure of the disproportion between the

number of positive and negative deflections which would help differentiate between

background and evoked activity, a parameter D (=Disproportionality) was defined as

the normalized difference between the numbers of positive and negative deflections

obtained in the particular time bin over the set of trials:

where

#p –number of positive deflections obtained in particular time bin, - obtained from the

Time Distribution of Positive Deflections;

#n –number of negative deflections obtained in particular time bin, - obtained from

the Time Distribution of Negative Deflections;

The D value should be distributed approximately as a standard normal

variance N(0,1) if the probability to obtain positive deflections in the given time bin

over N trials is equal to the probability of obtaining negative deflections (i.e.

background activity). Figure 10 shows the D-values obtained during the resting EEG

(absence of stimulus) and in the 300 ms pre-stimulus periods in a typical young

subject (subject SB), along with the distribution N(0,1) superimposed. This figure

illustrates that the experimental data during background activity is in agreement with

a standard normal variance distribution.

)#(#/)#(# npnpD

Page 49: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

49

Figure 10. Statistical Distribution of D-values during Background Activity. D-values during resting EEG (absence of stimulus) and during pre-stimulus period - subject SB (one session). X-axis : D-values; Y-axis: Percentage of D-values obtained in the given bin of the X-scale. Filled squares = 300 ms pre-stimulus period (background); Empty squares = resting EEG; Bold line = distribution of normal standard variance N(0,1).

Entering values of D into the table of normal (Gauss) distribution N(0,1)

provides the criterion for assessing the presence of a D value which is significantly

greater than that during background activity, i.e. the presence of a significant

disproportion between the number of positive and negative deflections. A greater

disproportion would give a higher D value, indicating the presence of substantially

more positive than negative deflections (or vice versa) over repeated stimulus trials,

which can be considered as time locking of deflections (which would then be evoked

activity). For example, a duration of 1300 ms (216 time bins of 6 ms) requires a

critical level Dcr (p <0.05) = 3.5. Note that the critical level of the D value is subject

independent.

Page 50: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

50

Onset, termination and duration of the overall evoked response in young and

elderly subjects

Figures 11A and 11B show the time distributions of the positive and of the

negative deflections and in addition, a plot of the D values derived from these time

distributions from all 14 young subjects, superimposed, and separately those for the

14 elderly subjects, for electrodes Oz (Figure 11A) and Pz (Figure 11B).

Page 51: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

51

Page 52: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

52

Figures 11A and 11B. Time distribution of positive deflections and of negative deflections in all 14 young (left) and in all 14 elderly (right) subjects, superimposed. Below are the respective D-values. The two horizontal lines in the graphs of the D values represent +3.5. X-axis: Time before and after the onset of the stimulus [ms]- time 0 = trigger of visual stimulus reversal. Y-axis: Percentage of deflections and D-values. The two horizontal lines show ±3.5 standard deviations. 11A – at electrode Oz; 11B – at electrode Pz.

Page 53: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

53

Using the 3.5 standard deviation D criterion (the critical level Dcr = 3.5 is

indicated in Figures 11A and 11B as the lines parallel to the time axis which are cut

by the profile of the D-values) for evaluating the time region of stimulus evoked

activity, the onset (defined as the time at which the number of the disproportionate

number of deflections of any polarity, and therefore the value of D, initially becomes

greater than the 3.5 standard deviations criterion), the termination (defined as the time

when the D value finally returns to background level before 300 ms post-stimulus)

and duration (the time between the onset and the termination) were determined for the

overall response in both age groups. One can see a significant increase in the D-value

in both directions (positive and negative) in a specific time range (e.g. from 50 to

about 250 ms, the period within which the conventional evoked potential is usually

obtained). The overall evoked activity begins earlier (shorter latency) and has a longer

duration in the young subjects. These data for electrodes Oz and Pz are presented in

Table 1.

Table 1. Mean (SD) of the Onset, Duration, and Termination of the Overall

Response in the Young and Elderly Groups at Oz and Pz electrodes using D value

of +3.5.

Electrode Oz Onset: [ms] * Duration: [ms] Termination: [ms]

YOUNG 63.5 15.3

153.2 37.9 213.7 33.0

ELDERLY 88.4 25.0

131.7 62.2 215.4 44.7

Electrode Pz Onset: [ms] Duration: [ms] * Termination: [ms] *

YOUNG 82.1 32.8

157.7 52.5 236.9 43.1

ELDERLY 87.0 18.4

108.6 48.9 190.7 37.1

* Significant Group Difference p < 0.05

Note that the onset of the response at electrode Oz in the young was

significantly earlier than in the elderly (young = 63.5 15.34 ms, elderly = 88.42 25

ms; p=0.0075, two-tail t-test) (see Table 1). This earlier onset of activity seems to be

Page 54: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

54

related to the presence of an additional positive peak in the time distribution of

positive deflections and in the D-values at approximately 70 ms that appeared in

almost all (N=10) young subjects (see Figure 11B), but only in one elderly subject.

This peak was not apparent neither in the amplitude profiles nor in the conventionally

averaged VEPs derived from the same data in most of the subjects (present in the

amplitude profiles in only 1 young and 2 elderly subjects; in the conventionally

averaged VEP in only 1 elderly subject). Also an analysis of electrode Pz showed

significant differences in young and elderly subjects in regards to the duration as well

as the termination of the response (duration: young = 157.7 52.5 ms, elderly = 108.6

48.9 ms; p=0.036) (termination: young = 236.9 43.1 ms, elderly = 190.7 37.1

ms; p=0.0138). The longer duration and later termination of the response seems to be

the result of the majority of the young subjects displaying an additional later positive

peak (best seen at Pz in the distribution of D-values), about 220 ms after the stimulus

(see Figure 11B). This later wave was seen in only a few of the elderly subjects. A

similar analysis can be conducted for each of the other recording electrodes.

Time distributions of the deflections during background activity in young and elderly

subjects - Rate of deflections

In order to assess whether there are differences in the mean number of

deflections of the same polarity during background activity between the young and

elderly subjects, the respective time distributions of the deflections were analyzed.

The analysis was conducted on the data obtained from all 6 sessions in each group for

all subjects for electrode Oz and revealed significant differences between groups in

the rate of background activity at electrode Oz, with the elderly displaying a higher

rate (mean 17.3 1.4 deflections/sec), compared to the young subjects (mean rate of

15.3 0.8 deflections/sec) (p < .001).

Page 55: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

55

The same analysis used for the Oz electrode can also be used for all electrodes.

A separate analysis across 16 electrode sites revealed significant group differences

with the elderly having a greater mean number of positive EEG deflections (F (1, 26)

= 13.84, p < .001) in the 300 ms pre-stimulus period. The number of deflections per

second of background activity in the 300 ms pre-stimulus periods for each of the 16

electrodes is shown in Figure 12. Although there were significant group differences

with regards to the rate of deflections, where the elderly had a higher frequency of

background activity, the pattern of these differences across electrode sites was similar

in both groups, indicating that there were no significant scalp topographic differences

between the young and elderly. Thus the rate of deflections in the 300 ms pre-stimulus

period (background activity) in both groups lies between 14-20 deflections/sec and is

significantly greater in the elderly compared to the young subjects.

Figure 12. The number of deflections/sec of background activity for 14 young and 14 elderly subjects across 16 electrodes. X-axis: electrodes. Y-axis: mean rate [deflections/sec].

Page 56: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

56

Amplitudes of the deflections during background activity in young and elderly

subjects

Analysis of electrode Oz revealed significant group differences in the base to

peak amplitudes of the same polarity during background activity with the young

displaying higher amplitudes (4.73 1.9 V), compared to the elderly subjects (3.70

0.9 V; p < .001). In fact there were significant differences in the amplitudes of

deflections in background activity for all three parietal as well as occipital electrodes

(Pz, P3, P4, Oz, O1, O2) and not at the other electrodes with larger amplitudes in the

young when compared to the elderly (See Figure 13).

Figure 13. The mean amplitudes of deflections in background activity for 14 young and 14

elderly subjects across 16 electrodes. X-axis: electrodes. Y-axis: V.

A significant Age Group x Electrode Location interaction for the amplitudes

of deflections was found during background activity (F (15, 390) = 10.95, p < .000).

In the young the smallest amplitudes were at electrode T4 (3.46 0.84 V) and largest

at electrode Pz (5.90 2.33 V), while the smallest amplitudes in the elderly were at

electrode Oz (3.70 .85 V) and largest at electrode F4 (5.85 2.35 V). Thus there

is a different scalp topographical pattern of the amplitudes of background deflections

between the young and the elderly.

Page 57: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

57

Relation between mean rate of EEG deflections during the 300 ms pre-stimulus

background activity and during the 300 ms overall evoked post-stimulus activity

The mean rate of deflections during the time period associated with

background activity (300 ms pre-stimulus period) was compared to that in the time

period containing all of the overall evoked response in all subjects [(50-350 ms post-

stimulus period (see Table 1)]. The analysis of the Oz-records revealed that while

young subjects showed no significant differences in mean rate between these two time

periods (young: background period = 15.25 .78 deflections/sec; evoked period =

15.10 .67 deflections/sec), the elderly displayed significant differences between the

two time periods, with a lower rate during the evoked period compared to the

background activity (elderly: background period = 17.11 1.52 deflections/sec,

evoked response = 16.75 1.78 deflections/sec (p < .02).

Time distributions of deflections related to the individual components of the

evoked response - Partial time distributions of deflections corresponding to the

components of the evoked response: determination and parameters used

The time period associated with the overall evoked response can be further

broken down into the time periods that correspond to the typical conventional

averaged evoked potential VEP components i.e.: N1, P1 and N2. The correspondence

between the known components of conventional EP, the time distributions of

deflections and D-values permits the separation of the overall time distribution of

deflections of evoked activity into its local distribution for individual EP components.

Each such time period is characterized by an increase in the number of deflections of

a specific polarity (over that of the opposite polarity) at a specific time period. For

example, with respect to the peak of N1, the point in time where the D value crosses

Page 58: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

58

the zero line in the negative direction (see Figure 11A) is defined as the beginning of

this partial time distribution of deflections (T1), while the ascending zero crossing is

the end (T2) of this distribution. In this way one can define a partial time distribution

of deflections for each component (i.e.: N1, P1 and N2). Thus the T2-T1 for each

wave can be considered the width or duration of a partial time distribution of

deflections for that wave (Note that here when determining the partial time

distributions for each component, a D-value of zero was used, while when assessing

the overall response, a D-value of +3.5 was used).

Borders and duration (width) of the partial time distributions related to the N1, P1

and N2 components of VEP in young and elderly subjects

The values of T1, T2 and T2-T1 at Oz electrode for the young and elderly

subjects are shown in Table 2. For component P1 at Oz electrode, the elderly subjects

demonstrated a significantly longer latency of T2 (p = 0.000072, two-tail t-test).

Significant group differences of T1 for component N2 (p = 0.0325) were also

obtained, where the young subjects had an earlier onset of the response related to that

partial time distribution of deflections. Table 2 shows that there is a significant aging

effect for T2-T1 (duration) of the partial time distributions of deflections for the

individual components N1 (p = 0.0285) and P1 (p = 0.0197), with the elderly having

longer durations of the distributions. The longer durations (T2-T1) for components N1

and P1 in the elderly subjects are signs that the deflections are more dispersed in time,

that there is poorer time locking and greater time variability. From the time

distributions and the plot of the D values in Figure 11A, one can see that the partial

time distributions of the N1 and P1 deflections are also more prominent and consistent

in the young subjects.

Page 59: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

59

Table 2. Borders and Duration (mean SD) of the Partial Time Distributions of

Deflections contributing to the N1, P1 and N2 components of the evoked activity in

the young and elderly groups at electrode Oz, using D value of zero.

T1 [ms]: Left border of the partial

distribution N1 P1 N2*

Young 73.4 7.4 103.8 5.3 133.6 4.3

Elderly 69.1 12.3 108.5 11.7 148.4 12.2

T2 [ms]: Right border of the partial

distribution N1 P1* N2

Young 94.8 6.5 128.2 4.6 171.0 17.4

Elderly 95.6 11.5 140.2 7.6 177.9 14.6

T2 – T1 [ms] Duration of the

Component N1* P1* N2

Young 24.4 3.0 27.4 4.8 40.4 16.5

Elderly 29.6 7.4 34.7 9.2 32.6 11.2

* Significant Group Differences p< 0.05

Mean number of deflections per single stimulus trial for the partial time distributions

of N1, P1, N2 in young and elderly subjects

The number of deflections appearing within the duration (T2-T1) of each

partial time distribution (for the N1, P1 and N2 components) over all trials (300) in a

session was assessed. Dividing this total number of deflections in a session by the

number of trials in the session (300), gives the mean number of deflections per

stimulus trial. Table 3 presents the data obtained for components N1, P1, N2 at

electrode Oz in young and elderly subjects. Note that these mean values of

deflections per trial are less than one, showing that not all stimulus trials contributed

deflections within the time period related to the given response component, i.e. there

are single stimulus trials which fail to contribute to the average response. One can see

that the mean number of deflections per stimulus trial related to components N1, P1

and N2 (consistently less than one) is not significantly different between young and

elderly subjects.

Page 60: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

60

Table 3. Mean number of deflections per trial contributing to Time Regions

(T1, T2) of the Response Components (Resp) and relative number of

contributing deflections with regards to background activity (Bgrd) in the

Young and Elderly. Electrode Oz

Component N1

Parameter Deflections/Trial Ratio: Defl(Resp)/Defl(Bgrd)*

Young 0.69 0.11 1.50 0.23

Elderly 0.64 0.14 1.19 0.10

Component P1

Parameter Deflections/Trial Ratio: Defl(Resp)/Defl(Bgrd)*

Young 0.79 0.10 1.59 0.26

Elderly 0.82 0.18 1.27 0.14

Component N2

Parameter Deflections/Trial Ratio: Defl(Resp)/Defl(Bgrd)

Young 0.86 0.23 1.28 0.13

Elderly 0.77 0.19 1.21 0.09

* Significant Group Differences p< 0.05

Only those stimulus trials with a deflection in the relevant time period of each

of the components N1, P1 and N2 could contribute to the corresponding component

found in the conventionally averaged VEPs derived from the same data. These can be

referred to as stimulus trials with appropriate deflections (responding trials) and can

be used to differentiate between contributing and non-contributing single trials and

lead to further separate analysis (single trial analysis). The onset and termination of

the partial time distributions of the deflections for each response component was

determined (Table 2) in each subject. Therefore one can go back and scan all single

stimulus trials in each subject and identify and extract those single stimulus trials

which had an appropriate deflection within the T1-T2 time period of the partial time

distribution of a component wave. This has been done for component P1 in all young

and elderly subjects and there was a mean number of 235.5 29.8 responding trials

out of the 300 in a session (i.e. 78.5%) in the young subjects and 239.2 41.4 in the

elderly (i.e. 79.7%). There was no significant difference between the mean numbers

of responding trials between the young and elderly subjects and the mean number

Page 61: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

61

across all subjects (young and elderly) which contributed to the P1 component was

79.1%. A similar analysis could be conducted for components N1 and N2.

One can also compare, during the partial time distribution of deflections of

each of the components (N1, P1, N2 separately), the ratio of the number of evoked

period deflections to the number of deflections which occurred during an equivalent

time duration during the pre-stimulus period over all trials in a session. This too is

shown in Table 3 (right). Note that the ratios for each of the response components

separately are consistently greater than one, indicating that there are more deflections

during the partial time distribution periods of evoked activity compared to equivalent

periods of background activity and the ratios are significantly different between the

young and the elderly for components N1 (p = .0001) and P1 (p = .0001), where the

young demonstrated a greater concentration of deflections in the time period related to

the response than during the equivalent time period of background activity.

Variability of latencies of the EEG deflections contributing to the N1, P1 and N2

components of VEP

The dispersion of the deflections within the partial time distribution of a

response component may differ between the young and elderly and the degree of this

dispersion (variability) can be expressed in several ways. For example in a previous

section, it was shown (Table 2) that in the elderly, the durations of the partial time

distribution of components N1 and P1 were longer (an indication of greater

dispersion, variability and poorer time locking) than those in the young.

The dispersion of the deflections contributing to each response component can

also be quantified as the standard deviation around the mean time of those deflections

contributing to each component. The mean latency of deflections contributing to each

Page 62: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

62

response component (N1, P1, N2) in both groups and its standard deviation are given

in Table 4 for electrode Oz. The mean latencies of deflections contributing to the

partial Time distributions related to N1, P1 and N2 components of the response at the

Oz electrode did not differ between the groups, but the standard deviations of the

latencies for electrode Oz were significantly smaller (less time jitter) in the young for

components N1 (p = .011) and P1 (p = .0001).

Table 4. The variability of the responses within the Partial Time Distribution of

Deflections for the young and elderly groups at electrode Oz.

Mode [ms] – time of the peak number

of deflections: Tpeak N1 P1 N2*

Young 84.3 8.7 116.5 4.4 152.5 11.0

Elderly 83.9 15.1 113.9 12.3 164.1 12.4

Mean Latency of deflections [ms] N1 P1 N2

Young 85.7 7.1 115.7 4.6 155.0 12.7

Elderly 82.8 11.8 121.7 9.8 163.6 12.4

Standard Deviation of the deflection’s

appearance: [ms] N1* P1* N2

Young 6.9 0.8 7.5 1.3 11.0 2.9

Elderly 8.7 2.2 11.5 2.7 9.7 3.1

Asymmetry (mode – mean) [ms] N1 P1* N2

Young -1.4 2.8 0.8 3.2 -2.5 13.1

Elderly 1.1 4.8 -7.8 11.3 0.5 5.2

* Significant Group Differences p< 0.05

Finally the shape of the partial time distribution of deflections for each

component may not be symmetrical around the time of the peak number of deflections

(mode). This asymmetry can be expressed as the difference in milliseconds between

the time of the mode and the time of the mean. This is also given in (Table 4) and it

can be seen that the elderly displayed a significantly greater amount of asymmetry for

the peak of the partial time distribution of component P1 (p= .002) suggesting a

tendency to prolongation of the P1-latency in elderly in spite of the fact that the mean

and mode measures of the latency of the deflections did not differ significantly

between the young and elderly subjects.

Page 63: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

63

Time locking of the EEG response deflections

The magnitude of the D-value can serve as a measure of the degree of time

locking of deflections contributing to a response component since a greater D-value

indicates a greater degree of grouping of the deflections within a narrow time period

(see Figure 11A). The peak D-values derived from the partial time distributions of

deflections in the two age groups were therefore compared. This data is presented in

Table 5 for electrode Oz.

Table 5. Peak values of the D-values (from formula 1) related to the N1, P1 and N2

components of the evoked response in the young and elderly subjects at electrode Oz.

N1* P1* N2

Young (n=14) -12.3 5.0 15.2 4.6 -8.7 3.3

Elderly (n=14) -5.3 2.6 8.1 3.5 -6.8 4.0

* Significant Group Differences p< 0.05

This analysis revealed that the young displayed a higher degree of time

locking at the Oz electrode for all three components and it was significant for N1 (p =

.00022; two-tail t-test) and for P1 (p = .00019). Figure 11A shows the distribution of

D-values for components N1, P1, and N2. One can see that they are higher, sharper

and narrower in the young subjects. One can conclude from this section that the

timing of the responses in the young subjects have less time variability by all

measures used.

Comparison between latencies of the peak D-values of the partial time distributions of

deflections and latencies of conventionally averaged VEPs

Comparison of the times of the peak D-values for each component (N1, P1,

N2) between the two groups (young and elderly), presented in upper part of Table 6,

showed significant differences only with respect to the N2 component (p = 0.0415,

two tail t-test), which was longer in the elderly.

Page 64: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

64

The analysis of latencies derived from the conventionally averaged VEPs did

not reveal significant differences in latencies of the components between the groups

(Table 6, below). Comparison between the times of the peak D-values and the

latencies of the conventional VEP (from the same data) showed that the latencies of

the D-value peaks were significantly longer than the conventionally averaged VEP for

the N1 component [p = 0.000032 in young (n=14) and p = 0.0015 in elderly (n=12);

two-tail, paired t-test]. The opposite latency difference was seen for N2 in young (p =

0.00074, two-tail, paired t-test) and in elderly (p = 0.0032) subjects. On the other

hand, the P1 latencies of the D- peak and P1 VEP were practically identical both in

young and in elderly subjects.

Table 6. Latencies of the peak D-values of the Partial Time Distributions of

Deflections and of conventionally averaged VEPs in young and elderly subjects,

electrode Oz.

Latency Peak D-Values [ms] N1 P1 N2*

Young 84.5 8.4 116.6 3.8 146.1 7.5

Elderly 86.2 13.1 122.2 12.0 160.4 12.0

Latency Conventionally Averaged

VEPs [ms]

N1 P1 N2

Young 79.6 8.7 116.1 4.8 160.4 14.3

Elderly 76.0 8.4 120.7 8.0 170.4 11.3

* Significant Group Differences p< 0.05

Amplitude amplification of the deflections contributing to the evoked response

components

The mean amplitude of the deflections which contributed to the partial time

distributions related to N1, P1 and N2 components can be considered as an additional

Page 65: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

65

characteristic of the processes leading to the generation of the EP in young and elderly

subjects. These were calculated by taking the mean of the amplitude profile of the

deflections which occurred within the T1-T2 time range specific for each component

and determined individually for each subject. These are given in the upper section of

Table 7. One can also define a parameter “amplification” which reflects the ratio of

this mean amplitude compared to the mean amplitude of the deflections during

background activity (300 ms pre-stimulus activity taken from Figure 12). Such data in

young and elderly subjects at the Oz electrode are also presented in Table 7.

Table 7. Mean amplitude of the evoked period deflections (above), their ratios

compared to the background period (middle) and amplitudes of the conventionally

averaged VEPs related to the N1, P1 and N2 components, in the young and elderly

subjects. Electrode Oz.

Mean Amplitude of evoked period

Deflections V]

N1* P1* N2

Young -5.85 2.88 8.50 2.87 -5.32 2.06

Elderly -3.45 1.10 6.41 1.67 -4.54 1.65

Relative Amplitude (with regard to

background – pre-stimulus – activity)

N1* P1 N2

Young 1.17 0.22 1.88 0.52 1.11 0.44

Elderly 0.86 0.15 1.78 0.52 1.18 0.49

Amplitudes of the conventionally

averaged VEP (all trials) [V]

N1* P1* N2

Young -2.59 1.86 5.49 2.69 -2.25 2.08

Elderly -0.11 0.56 3.32 1.76 -1.52 1.88

* Significant Group Differences p< 0.05.

Note that the mean amplitude of the evoked period deflections is significantly

(2 tail t- test) larger in the young for N1 (p = 0.0028) and P1 (p = 0.0336)

Page 66: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

66

components. The ratio of evoked period to background amplitudes was significantly

larger in the young subjects only for the N1 component (p = 0.00044). In most cases

these amplitude ratios were greater than one, and significantly so only for P1 [both in

young (p = 0.0000076) and in elderly (p = 0.000053) subjects], and for N1 component

in young subjects (p = 0.028). This is an indication of an amplification of the evoked

period deflections over those during the background activity. Note that the degree of

this amplification for P1 did not differ between the groups. Even though there was a

significant degree of amplification with respect to N1 in the young subjects (ratio

1.17), the opposite phenomenon was observed in elderly group: this ratio (mean value

0.86) was significantly less than one (p = 0.0043, two-tail t-test). Thus the mean value

of the base-to-peak amplitude of evoked period deflections contributing to N1

component was even decreased (in relation to the background activity) in elderly

subjects, opposite to that seen in the young subjects. The mean amplitude of

deflections contributing to N2 did not differ significantly from background activity

(ratio not different from one).

The comparison of the amplitudes of the conventional VEP between young

and elderly showed the same results as the mean amplitude of the evoked period

deflections: there were significant differences for N1 (p = 0.00032, two-tail t-test)

and for P1 (0.034, one-tail t-test), and this difference for N2 was not significant.

One can also compare the mean amplitude of the evoked period deflections for

each component to the peak amplitude of the conventionally averaged VEP wave

component obtained from the same data (see Table 7). Note the significantly lower

amplitudes of the conventional averaged VEP compared to the corresponding values

of the mean amplitude of deflections. This is due to the finding that there is time jitter

in the response deflections making up the conventional VEP and that not all stimulus

Page 67: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

67

trials gave rise to appropriate deflections, providing evidence that conventional

averaging leads to a loss of information, including a smearing of amplitudes.

In addition, linear correlation analysis was conducted between mean amplitude

of deflections contributing to the EP components (N1, P1, N2) and mean amplitude of

the background deflections (see Figure 14). A significant correlation was found for all

three components in young subjects: N1 (r = 0.85, p=0.000092; coefficient of

regression 1.16, intercept 0.02), P1 (r = 0.76, p=0.0017; coefficient of regression 1.22,

intercept 2.75), N2 (r = 0.55, p=0.040; coefficient of regression 0.68, intercept 1.90).

In the elderly group a significant correlation between evoked and background

amplitudes was found only for the N1 component (r = 0.86, p=0.000077; coefficient

of regression 1.03, intercept -0.64). The range of background amplitudes in young

subjects (2.17 – 8.23 V) was significantly (p = 0.0044, one-tail F-test) wider than in

elderly (2.33 – 5.78 V). The correlation in the elderly subjects between background

and evoked period amplitudes for P1 and N2 were small and not significant.

Page 68: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

68

Figure 14. Scatter plot of mean amplitude of deflections during evoked activity [A: N1, B: P1, C: N2 components] on the Y-axis versus background amplitude (X-axis) in young (n=14) and elderly (n=14) subjects. Filled squares: young subjects; empty squares: elderly subjects. The linear regression line is for the young subjects (r values are presented; significant (p<0.05) correlation marked by *).

Page 69: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

69

Summary of Results and Discussion

In recent studies substantial attention has been devoted to the interrelation of

brain activity evoked by various stimuli and the background brain activity, which are

considered as being interconnected processes (Woody, 1967; Basar et al. 1997; Jansen

and Brandt, 1991; Rahn and Basar, 1993; Arieli et al., 1996; Kisley and Gerstein,

1999). The present study introduces several basic techniques for analysis of scalp

recorded background and evoked electrical activity based on the examination of the

number and of the amplitude of the positive and negative deflections of the EEG.

The chief results of this study include:

1. New analysis techniques provide a description of the evoked and

background EEG activity which then leads to statistical criteria and for the

quantitative differentiation between the background and evoked activity,

which revealed an interrelation of brain activity evoked by various stimuli

and background brain activity.

2. Analysis methods also provide an evaluation of temporal properties of

variation in EP activity (not accessible in conventionally averaged EP).

3. A higher rate of background activity in the elderly subjects when compared

to the young across all 16 electrode sites. The amplitudes of the deflections

during the pre-stimulus (background) period were larger in the young

subjects than in the elderly at the parietal and occipital electrodes.

4. In the elderly subjects there was a lower rate of deflections during the

period of evoked activity (50-350 ms after the stimulus) compared to that

during the 300 ms pre-stimulus (background) period. There was no such

difference in rates of deflections in the young subjects.

5. The critical value of the disproportionality of deflections (D) (3.5 ±

standard deviations) that was used to define the overall period of evoked

Page 70: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

70

activity was based on a calculation that is dependent on the overall time

interval used for analysis (i.e. 1300 ms) and on the number of bins in our

analysis (i.e. 216 - 6 ms bins). For example, if the recording period were

shorter thus containing fewer 6 ms bins, a smaller critical value would be

sufficient to determine the disproportionality of deflections related to the

stimulus evoked activity. The reverse would be true with a longer recording

period and greater number of bins. The overall period of evoked activity at

electrode Oz began significantly earlier in the young subjects (see Table 1)

and this seemed to be due to the presence of an earlier positive peak at about

70 ms. That was not seen in the elderly.

6. These new methods of analysis allow the generation of two subsets of

stimulus trials: those that contributed deflections to the partial time

distribution of for example: component P1 and those that do not. Therefore

we can study the properties of these two subsets separately, which is a form

of Single Trial Analysis. The mean number of trials across all subjects

(young and elderly) which contributed to the P1 component was 79%.

Again not all stimulus trials generated an appropriate response. There was

no significant difference between the mean numbers of responding trials (or

deflections) between the young and elderly subjects.

7. Analysis of three VEP components (N1, P1 and N2) using partial Time

distributions of deflections revealed significantly greater variability in the

elderly for components N1 and P1, as well as significantly higher standard

deviations of mean latency for all three components (N1, P1 and N2). The

reverse was true for the young, who demonstrated a greater degree of

stimulus time locking, and narrower time regions related to the VEP

components N1, P1 and N2 at electrode Oz. Greater variability and poorer

Page 71: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

71

time locking seen in the elderly may be indicative of age-related peripheral

visual and central deterioration, aging induced variance (jitter) in neuronal

conduction velocity and synaptic transmission.

8. VEP components N1, P1 and N2 are generated both by time-locking

(reorganization of time of deflections) and amplification of amplitude of

EEG deflections and these represent possible mechanisms involved in

generation of evoked activity.

9. The young had significantly higher mean amplitudes for the deflections

within the Amplitude Profiles as well as peak amplitudes from conventional

averages of the same data, related to components N1 and P1 than in the

elderly. Amplitudes measured using the Amplitude Profiles were

significantly higher than the amplitudes found in the conventional averages

derived from the same data, illustrating the distortion caused by averaging.

10. The ratios of the peak amplitudes of the Amplitude Profiles compared to the

mean amplitudes during background activity (pre-stimulus) for all three

response components were greater than one, indicating that there is

amplification during the evoked period. This degree of amplification (ratio)

was larger in the young subjects for N1 and P1 and the difference was

significant for wave N1.

The main result of the present study is the development of new techniques of EEG

and EP analysis which preserves the moment-to-moment variation in scalp recorded

activity, helps differentiate between evoked and ongoing (background) electrical

activity and provides a constructive basis for the study of single evoked responses.

Furthermore, the variability of the single responses can be evaluated by this method.

Page 72: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

72

In addition to the new information that can be gained using the Time

Distributions of Deflections and the Amplitude Profiles as well as the new method of

single trial analysis using two subsets, an important aspect of this work concerning the

new methods of analysis proposed here is that they will provide a less cumbersome

and a more simplified approach that can be used for the analysis of single trial evoked

potentials which provides an alternative to previous methodologies that have been

suggested. Some however may consider that the paradigm incorporated in this study,

involving considerable recording time (10 min) as well as many recording sessions (6

sessions – 3 per day) which were conducted over two different non-consecutive days,

to be unwieldy. However, the study was initially designed in this fashion due to the

fact that this was the first application and investigation of the newly proposed analysis

techniques, thus requiring that sufficient controls (i.e. more than one recording) be

conducted to ensure reliability and consistency of the data and findings over more

than one session and over different days, as well as being able to assess any possible

physiological or psychological changes an individual might experience such as

changes in mood or alertness that might induce variations in the recorded activity. It

was seen that the data collected over 6 recording sessions were internally consistent

(see Figure 9) and therefore there is no need to continue using 6 recording sessions. In

fact, visual inspection of the data (in Figure 6) suggested that a minimum of 100

stimulus trials from one recording session would be sufficient to obtain representative

time distributions of deflections and a minimum of 300 trials from 1 recording session

for representative amplitude profiles of the background and evoked activity for each

individual subject within the framework of this paradigm. In future studies fewer

recordings with shorter time durations of sessions should provide adequate data that

would reliability illustrate a representative picture of the subjects electro-cortical

activity. It is important to note that with the use of different types of stimuli (i.e. with

Page 73: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

73

different intensities or even pattern-reversal checkerboards with different check sizes,

different reversal rates, different modalities (e.g. auditory), different response

components (e.g. P50 of the AEP), etc…) as well as different experimental factors

(i.e. different subject groups or recording conditions), the number of trials or even the

number of sessions necessary to attain representative data using these proposed

analysis techniques might vary.

The algorithm used in the present study is based mainly on the relatively high

frequency range of the EEG (7-30Hz i.e. alpha and beta range). The high frequency

cutoff at 30 Hz is due to filtering of the EEG. The low frequency cutoff at 7 Hz is a

result of the algorithm, which detects all successive EEG deflections even if they are

small in amplitude, ignoring the low frequency EEG fluctuations upon which the

smaller amplitude, higher frequency deflections are superimposed. Synchronization

of the alpha-beta frequency deflections with onset of the visual stimulus seems to be

limited to latencies up to 200 ms. Thus later components (e.g. P300) might be

generated by EEG waves still lower in frequency content. In principle, the method

described here based on the time distributions of deflections and amplitude profiles

can be easily adapted to these later lower frequency components, by utilizing a filter

with a low frequency cutoff (below 7 Hz) and a slightly different algorithm.

These new analysis techniques, based on the time distributions of

deflections, D-values and amplitude profiles of deflections before (background)

and after (evoked) stimuli has succeeded in demonstrating several important

differences between young and elderly normal subjects.

Since the same new techniques were used here to analyze both background

and evoked activity, the properties of these two periods can be compared, e.g. rate of

deflections, their exact timing, and their amplitudes. A higher rate of background

activity was found in the elderly subjects compared to the young across all 16

Page 74: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

74

electrode sites (see Figure 12). These findings seem to be in agreement with previous

reports of higher Beta range frequencies in the elderly when compared to the young

(Polich and Luckritz, 1995; Polich, 1997), though different frequency analysis

techniques were used. The reason for this shift in Beta range frequencies in the

elderly is not clear.

The amplitudes of the deflections during the pre-stimulus (background) period

were larger in the young subjects than in the elderly at the parietal and occipital

electrodes (see Figure 13).

In the elderly subjects there was a lower rate of deflections during the overall

period of evoked activity (50-350 ms after the stimulus) compared to that during the

300 ms pre-stimulus (background) period. There was no such difference in rates of

deflections in the young subjects.

The overall period of evoked activity at electrode Oz, as studied in the time

distribution of the deflections and in the D values, began significantly earlier in the

young subjects and ended later (see Table 1) and this seemed to be due to the presence

in most of the young of an earlier positive peak at about 70 ms and a later peak at

about 220 ms (See Figure 11A). The earlier peak seen in young subjects may be

analogous to the short latency VEP previously reported (Pratt et al., 1982) in response

to flash stimuli. This earlier P70 peak seen in time distribution of deflections and D-

Values in the young subjects was not seen in the conventional averages derived from

the same data. This may be related to the fact that unlike the conventional VEP

which averages both the time and the amplitude of waves, the analysis of D-values

only considers timing. Thus, it is possible that since the P70 wave is smaller in

amplitude than the later N1, P1 and N2 waves and it may have greater amplitude

variability, it may not be readily apparent in conventional averages. It is also possible

that components N1 P1 and N2 undergo greater amplification than the P70

Page 75: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

75

component. Also the time distribution of deflections and D-Values emphasize activity

in a higher frequency range (Beta) with less representation of activity in lower

frequency ranges such as theta, delta, and alpha activity. In contrast, the conventional

averages take into account all four frequency ranges. Thus it is possible that the P70

component is generated more by Beta activity than are components N1, P1 and N2.

All of these factors together may contribute to the finding that the P70 is not seen

clearly in conventionally averaged data. Further investigation is necessary to analyze

these factors and as to why the P70 component was not seen in the elderly.

The longer duration and later termination of the overall period of response that

was seen in the young but not in all the elderly subjects was due to a later peak at

approximately 220 ms. This later peak may be related to a component that has been

referred to as the P2 component. Since it is thought that the longer latency

components may be related to cognitive processes, it is possible that the P2

component may reflect more higher order processing since the subjects were

instructed to count the auditory clicks. Thus the absence of the P2 component in the

elderly maybe due to age-related cognitive decline.

An additional interesting result which did not differ between young and

elderly subjects was the finding that only about 79% of the single stimulus trials

contributed positive deflections within the time period of component P1. Thus the

single stimulus trials in a session (300) can be broken down into two subsets of trials,

subset one which contributed positive deflections during the partial time distribution

of component P1 and subset two, which did not contribute such deflections. A

detailed comparison of the specific properties (time distributions, D-values, amplitude

profiles and averaged VEPs) of the two subsets of trials (those which contributed

deflections, separately from those which did not) during background and evoked

activity may contribute to an understanding of the reasons for the absence of

Page 76: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

76

appropriate deflections in the second subset. Finally, the lack of age-related

differences in relation to the percentage of stimulus trials that evoked a P1 response is

noteworthy considering the fact that differences were found with respect to other

response parameters and to the degenerative changes in the elderly when compared to

young. This may possibly suggest that the generation of appropriate deflections

related to a particular component (i.e. P1) may be less affected by age-related

neuropathological changes and even remain intact in old age. This requires further

investigation.

Also an interesting finding in the present study was the greater time variability

seen in the elderly in most measures used in this study: duration of a partial time

distribution, the standard deviation around the mean time of each component,

asymmetry between the mode and the mean of the partial time distribution, and the

magnitude of the D-value. This increased amount of variability in the elderly is a sign

of poorer time locking and may partially explain the significant reductions in

conventional averaged EP amplitudes in the elderly.

Of all the response parameters in this study (such as the Time distribution of

deflection, D-Values, Amplitude Profiles, amplification, and variance), only the

latency and amplitudes of the conventional averages derived from the same data can

be directly compared to studies in the literature, since they used only conventional

averaging. We will now compare findings of components N1, P1 and N2 to the

comparable findings in the literature.

Latency Measures

N1

In our study there were no significant age-related differences for the latencies

of component N1 from conventional averages (and for the latencies of the peak D-

values derived from the same data). There are varying reports of age related changes

Page 77: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

77

regarding the N1 wave. In a study by Allison et al. (1984) that tested PREPs in 286

healthy individuals aged 4 to 95 years of age, the authors report increasing latency

over the entire age range for the N1 wave. Other authors have also reported longer

latencies in the elderly for the N1 wave when compared to young subjects (Celesia

and Daly, 1977; Kjaer, 1980; Snyder et al., 1981; Shaw, 1984; La Marche et al., 1986;

Emmerson et al., 1994). Celesia et al. (1987) who conducted separate recordings of

PREPs on 112 normal individuals aged 20-70 with two different check sizes 15’ and

31’, found increased N1 latency in the elderly when using the 15’ check size, but in

the recordings in response to the 31’ checkerboards, they did not find any significant

age related differences for the N1 wave. In the present study we used a relatively large

check size (81’) since we thought that it would facilitate viewing in the elderly

subjects that are susceptible to age-related degeneration in the retina. Considering the

findings reported by Celesia et al. (1987) and the fact that we used a relatively large

check size, perhaps the lack of age-related differences is an indication that as the

check size increases the pattern-reversal stimulus becomes less sensitive to latency

differences between young and elderly subjects.

Although there were no significant latency differences for the peak D-values

and for latencies from conventional averages from the same data for component N1

between the two groups in our study, the elderly showed a trend for earlier latencies

than those in the young with respect to latencies derived from conventional averages

(see Table 6), and this may be related to findings reported by Kugler (1999), who

recorded PREPs for a group of 289 healthy individuals ages 18-98 years and found

that N1 latencies were actually earlier for the elderly group in comparison to the

young group.

An additional finding in this study related to latency measures is the

significantly longer latencies of the peak D-values of N1 compared to those found in

Page 78: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

78

the conventionally averaged VEP from the same data, both in young and elderly

groups (see Table 6). This discrepancy may partially be due to the conventional

averaging procedure which includes all trials, while the partial time distributions is

composed only of trials which contributed deflections to the particular components

being analyzed (see discussion of 2 subsets above).

P1

In this study no P1 latency differences were found in the conventional

averages between the two groups in this study. [The same was true for the latencies of

the peak D-values derived from the same data]. This is in disagreement with the

widely reported finding in studies that the pattern reversal VEP P1 component of

conventionally averaged data shows a significant increase in latency after the 6th and

7th decades of life (Asselman et al., 1975; Celesia and Daly, 1977; Kjaer 1980; Shaw

and Cant, 1980; Snyder et al., 1981; Sokol et al., 1981; Allison et al., 1984; Kazis et

al., 1983; Allison et al., 1984; Shaw, 1984; Verma and Kooi, 1984; Tobimatsu et al.,

1993; Emmerson et al., 1994). One notable exception to these findings was reported

by Celesia et al. (1987) who, as mentioned above, conducted separate recordings of

PREPs with two different check sizes 15’ and 31’. The 15’ checkerboards revealed

the increased P1 latency in the elderly, but the recordings using the 31’ checkerboards

again did not elicit any significant differences of P1 latency between young and

elderly. Thus the absence of P1 latency differences in our study may be due to the

large check size (81’) which we used (see discussion regarding N1 findings). Another

notable exception was reported by Kugler (1999) who found that the P1 VEP

component did not reveal any significant age related changes in latency. [Thus our

results with respect to the peak times of the D-values (no difference between young

and elderly) are in agreement with these latter studies].

Page 79: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

79

When comparing the latencies of the peak D-values with the latencies of the

conventionally averaged VEP for the P1 component in our results, the latencies were

roughly the same for both latency measures, both in young and elderly groups (see

Table 6) (also see discussion regarding related N1 findings).

N2

There were no differences with respect to the latency of the conventionally

average N2 VEP component between the young and elderly in our study. Snyder et al.

(1981) also reported that there were no N2 latency differences between the young and

elderly groups in their study. However, Kjaer (1980) reported that the latency of N2

(referred to as N135) in the elderly was shorter than that in the young group. In

contrast to these findings, there have been reports of increased latency after the 6th

decade for component N2 (Allison et al. 1984; La Marche at al. 1986; Shaw, 1984 and

Sidman et al. 1991).

An additional characteristic of the N2 component was found when evaluating

the partial time distribution for N2, revealing that several elderly subjects displayed

not only one peak in the N2 time range but often two peaks at slightly different

latencies (this can be seen in Figures 11A and 11B). The first N2 peak that appeared

in many of the elderly (145 ms) often had approximately the same latency as the

single N2 peak in the young (146 ms). A second N2 peak appeared in many of the

elderly at 160 ms. It is important to note that the elderly that displayed only a single

N2 peak, had longer latencies (160 ms) than the single N2 peak in the young (146

ms). This suggests that the latency of the conventionally averaged N2 component may

in fact be the result of the averaging together of two sub-components (i.e. N2a and

N2b), which then gives a latency value similar to that in young subjects. Kjaer (1980)

also reported that the N2 was the most variable component in the waveform within a

Page 80: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

80

normal group of young and elderly subjects, thus having to omit its latencies from the

analysis.

Similar to studies that reported increased latency of the conventionally

averaged N2 in the elderly (Allison et al. 1984; La Marche at al. 1986; Shaw, 1984

and Sidman et al. 1991), there were in the present study significant age-related

differences with respect to the latencies of peak D-values, with the elderly having

longer D-value latencies. Since not all elderly had the earlier peak, only the latter peak

was included in the comparison of the latencies of peak D-values in the present study,

thus resulting in significantly longer D-value latencies in the elderly (see Table 6).

Again when we compared the latencies of the peak D-values with the latencies

of the conventionally average VEP for the N2 component, the latencies for the peak

D-values were significantly shorter than those found in the conventionally averaged

VEP both in young and elderly groups (see Table 6) (also see discussion regarding

related N1 findings).

When longer N1, P1 or N2 latencies were found in the elderly, they have been

explained by the retinal and neuronal changes mentioned previously. Not all workers

have found such longer latencies in the elderly. Shaw and Cant (1980) have suggested

that the discrepancy between the published results may be partly accounted for by

different pattern luminances employed, as well as other differences in other stimulus

parameters. For example, as discussed earlier, Celesia et al. (1987) found that P1

latency was increased in elderly with smaller check sizes used in PREPs. This was not

seen when larger check sizes were used. This suggests a differential effect of aging on

the various spatial frequency channels of the visual system.

Page 81: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

81

Amplitude Measures

The amplitudes of VEPs in the young and elderly have also been widely

studied, but amplitude measures are far more variable than findings concerning P1

latency. In the present study the amplitudes of components N1 and P1 derived from

the conventional VEP from the Oz electrode were significantly greater in the young

than the elderly (see Table 7), as has also been previously reported for the pattern

reversal VEP components N1 in several studies (Kjaer, 1980; Sidman et al. 1991) and

P1 (Kjaer, 1980; Shaw and Cant; 1981). In contrast to these, no significant age related

amplitude changes were found in other studies for the amplitudes of components N1

(Celesia and Daly, 1977; Allison et al., 1984; La Marche et al., 1986; Celesia et al.,

1987; Emmerson et al., 1994) and P1 (Celesia and Daly, 1977; Kazis et al., 1983;

Allison et al., 1984; Verma and Kooi, 1984; La Marche et al., 1986; Celesia et al.,

1987; Sidman et al. 1991; Tobimatsu et al., 1993; Emmerson et al., 1994; Kugler,

1999) between young and elderly. These discrepancies with respect to VEP

amplitudes may be due to different sample sizes and experimental designs which

could yield different amplitude findings regarding age, since amplitude measures are

generally more variable. There were no group differences in amplitudes of component

N2 derived from the conventional VEP from the Oz electrode. Previous findings also

support the lack of any notable group differences with regards to the VEP N2

amplitude (Allison et al., 1984; La Marche et al. 1986), as was seen in this study.

In the present study, significant group differences were also found with respect

to the mean amplitudes of components N1 and P1 derived from the amplitude

profiles, which were greater in the young than the elderly (see Table 7). No group

differences were found with respect to mean amplitudes of component N2 derived

from the amplitude profiles.

Page 82: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

82

Amplitudes measured using the amplitude profiles were significantly greater

than the amplitudes found in the conventional averages derived from the same data,

illustrating the distortion caused by averaging (see Table 7). This is probably due to

the fact that the conventional averaging procedure makes use of all stimulus trials in a

session, including those which did not contribute deflections to the response, while

the amplitude profile, by definition is made up only of those deflections which

occurred within the relevant partial time distributions of each component. Thus the

conventional averages are “diluted” by non-contributing trials. In addition, time jitter

of deflections also causes a reduction in conventional averaged EP amplitudes. This

finding highlights one of the main advantages of these new analysis techniques.

The ratios of the mean amplitudes of the amplitude profiles (evoked period)

compared to the mean amplitudes during background activity (pre-stimulus) for P1

component were greater than one, both in young and elderly subjects, indicating that

there is an amplification during the evoked period. There was no difference in the

degree of the P1 amplification (ratio) between young and elderly subjects. Since the

mean P1 amplitude in the young was significantly greater than that in elderly, the

finding of similar amplitude ratios for P1 in young and elderly, is probably due to the

smaller amplitudes of background activity in the elderly at the Oz electrode (see

Figure 13). A significant amplification was obtained also for the N1 component in

young subjects, but the opposite phenomenon (ratio was significantly less than one)

was obtained in elderly subjects (see Table 7).

There was also a significant positive dependence (positive linear correlation)

between evoked period deflection amplitudes for all components (N1, P1, N2) and

background amplitudes in young subjects, and for component N1 in elderly also. This

suggests a strong connection between background activity and evoked activity (e.g. if

Page 83: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

83

there is low amplitude background activity then there will also be a lower level of

evoked activity amplitude and vice-versa).

When considering the overall results from the time distributions of deflections

and the amplitude profiles, it is likely that the longer latencies, reduced amplitudes

and greater time variability seen in the elderly are due to age-related changes, for

example in the retina (Porceddu et al., 1990), optic nerve (Dolman et al., 1980), to

decreased nerve conduction velocities (Dorfman and Bosley, 1979), to decreases in

the number of synapses per unit volume of brain tissue (Bertoni-Freddari et al., 1996),

to white matter atrophy (Meier-Ruge et al., 1992), to reduction in neurotransmitters

(Allen et al., 1983) and to decreases in myelin (Lintl and Braak, 1983). These

anatomical and functional changes seen during aging would induce greater variance

(jitter) in neuronal conduction velocity and synaptic transmission, with less consistent

signal processing, a greater variability and longer latencies. On the other hand the

young have intact conduction and cortical processing leading to greater consistency

with less variability in response to the stimuli (i.e. time locking). Topographical

analysis of differences in latencies in the partial time distribution of the various

components, D-values, and mean amplitudes of the Amplitude Profiles may also

provide additional information regarding age-related cortical changes associated with

visual EPs.

In final analysis it seems that the appearance of a peak (component) in the

conventional averaged EP in response to repeated stimuli is probably determined by

two main mechanisms: the time locking of the deflections and an increase in the

amplitude of these deflections (amplification - i.e. the amplitude of the deflections

during the specific period is greater than that during background activity). This latter

mechanism may be attributed to enhanced synchronization of synaptic-dendritic

Page 84: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

84

potentials in the cortex initiated by the stimuli. The degree of possible influences of

the amount of time locking of deflections and their amplification on the final EP

should be investigated in further studies. These new techniques can also be used in the

future to assess whether habituation occurred during the recording, and at what point

in time during the session.

The novel methods presented here provide a great deal of additional

information pertaining to age-related differences between young and elderly. This is

with respect to both cortical activity directly related to the response as well as

background activity. This additional information is unavailable when analyzing data

using only conventional averaging or spectral analysis techniques. These novel

techniques provide accurate and reliable additions to conventional averaging, so that

background and evoked activity can now be described also by using the time

distributions of deflections, D-values and amplitude profiles with less loss or

distortion of data. It also contributes to single trial analysis. The conventional EP

averaging technique can also be improved by using only those stimulus trials which

contribute deflections to the various response components. These techniques have also

clearly demonstrated a strong dependence of evoked period activity on background

activity, with respect to timing and amplitude. Therefore it is likely that these

techniques will allow researchers to access important information concerning

underlying physiological differences in a wide range of groups as well as various

types of neuropathology that would not otherwise be possible using the more

conventional forms of analysis.

Page 85: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

85

References

Allen SJ, Benton JS, Goodhardt MJ, Haan EA, Sims NR, Smith CC, Spillane JA,

Bowen DM, Davison AN. Biochemical evidence of selective nerve cell changes in the

normal ageing human and rat brain. J Neurochem 1983;41:256-265.

Allison T, Hume AL, Wood CC, Goff WR. Developmental and aging changes in

somatosensory, auditory and visual evoked potentials. Electroencephalogr Clin

Neurophysiol 1984;58:14-24.

Arieli A, Sterkin A, Grinvald A, Aertsen A. Dynamics of ongoing activity:

explanation of the large variability in evoked cortical responses. Science

1996;273:1868-1871.

Asselman P, Chadwick DW, Marsden DC. Visual evoked responses in the diagnosis

and management of patients suspected of multiple sclerosis. Brain 1975;98:261-282.

Basar E, Rahn E, Demiralp T, Schurmann M. Spontaneous EEG theta activity

controls frontal visual evoked potential amplitudes. Electroencephalogr Clin

Neurophysiol 1998;108:101-109.

Basar E, Schurmann M, Basar-Eroglu C, Karakas S. Alpha oscillations in brain

functioning: an integrative theory. Int J Psychophysiol 1997;26:5-29.

Bertoni-Freddari C, Fattoretti P, Paoloni R, Caselli U, Galeazzi L, Meier-Ruge W.

Synaptic structural dynamics and aging. Gerontology 1996;42:170-180.

Braak H, Braak E. Morphology of the human isocortex in young and aged individuals:

qualitive and quantitative findings. In: Ulrich J, editor. Histology and histopathology

of the aging brain. Basel: Karger S, 1988:1-15.

Buchthal F, Rosenfalck A. Evoked action potentials and conduction velocity in human

sensory nerves. Brain Res 1966;3:1-122.

Celesia GG, Archer CR, Kuroiwa Y, Goldfader PR. Visual function of the

extrageniculo-calcarine system in man: relationship to cortical blindness. Arch Neurol

1980;37:704-706.

Celesia GG, Daly RF. Effects of aging on visual evoked responses. Arch Neurol

1977;34:403-407.

Celesia GG, Kaufman D, Cone S. Effects of age and sex on pattern electroretinograms

and visual evoked potentials. Electroencephalogr Clin Neurophysiol 1987;68:161-

171.

Cobb WA, Dawson GD. The latency and form in man of the occipital potentials

evoked by bright flashes. J Physiol 1960;152:108-121.

Dolman CL, McCormick AQ, Drance SM. Aging of the optic nerve. Arch

Ophthalmol 1980;98:2053-2058.

Dorfman LJ, Bosley TM. Age-related changes in peripheral and central nerve

conduction in man. Neurology 1979;29:38-44.

Page 86: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

86

Downie AW, Newell DJ. Sensory nerve conduction in patients with diabetes mellitus

and controls. Neurology 1961;11:876-882.

Drechsler F. Quantitative analysis of neurophysiological processes of the aging CNS.

J Neurol 1978;218:197-213.

Dustman RE, Shearer DE, Emmerson RY. EEG and event-related potentials in normal

aging. Prog Neurobiol 1993;41:369-401.

Emmerson-Hanover R, Shearer DE, Creel DJ, Dustman RE. Pattern reversal evoked

potentials: gender differences and age-related changes in amplitude and latency.

Electroencephalogr Clin Neurophysiol 1994;92:93-101.

Ford JM, Pfefferbaum A. Event-related potentials and eyeblink responses in

automatic and controlled processing: effects of age. Electroencephalogr Clin

Neurophysiol 1991;78:361-377.

Ford JM, White P, Lim KO, Pfefferbaum A. Schizophrenics have fewer and smaller

P300s: a single-trial analysis. Biol Psychiatry 1994;35:96-103.

Giaquinto S. Aging and the nervous system. Chichester: Wiley, 1988: 34-69.

Gartner S, Henkind P. Aging and degeneration of the human macula: outer nuclear

layer and photoreceptors. Brit J Ophthalmol 1981;65:23-28.

Giacobini E. Cholinergic Receptors in human brain: effects of aging and Alzheimer

disease. J Neurosci Res 1990;27:548-560.

Halliday AM, Michael WF. Changes in pattern-evoked responses in man associated

with the vertical and horizontal meridians of the visual field. J Physiol 1970;208:499-

513.

Hillyard SA, Picton TW. Electrophysiology of cognition. In: Mountcastle VB et al.,

editor. Handbook of Physiology Vol. V: Higher functions of the brain, Part 2.

Baltimore: American Physiology Society, 1987: 519-583.

Hubel DH, Wiesel TN. Receptive fields of single neurones in the cat's striate cortex. J

Physiol 1959;148:574-591.

Hubel DH, Wiesel TN. Receptive fields and functional architecture of monkey striate

cortex . J Physiol 1968;195:215-243.

Hubel DH, Wiesel TN. Sequence regularity and geometry of orientation columns in

the monkey striate cortex. J Comp Neurol 1974;158:267-293.

Iwangoff P, Armbruster R, Enz A, Meier-Ruge W. Glycolytic enzymes from human

autoptic brain cortex: Normal aged and demented cases. Mech Aging Dev 1980;14:

203-209.

Jansen BH, Brandt ME. The effect of the phase of prestimulus alpha activity on the

averaged visual evoked response. Electroencephalogr Clin Neurophysiol

1991;80:241-250.

Page 87: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

87

Jeffreys DA, Axford JG. Source locations of pattern-specific components of human

visual evoked potentials: I. Component of striate cortical origin. Exp Brain Res

1972;16:1-21.

Kazis A, Vlaikidis P, Pappa P, Papanastasiou J, Vlahveis G, Routsonis K.

Somatosensory and visual evoked potentials in human aging. Electromyogr Clin

Neurophysiol 1983;23:49-59.

Kiloh LG, McComas AJ, Osselton JW, Upton ARM. Clinical

Electroencephalography. London: Butterworths, 1981:1-80.

Kisley MA, Gerstein GL. Trial-to-trial variability and state-dependent modulation of

auditory-evoked responses in cortex. J Neurosci 1999;19:10451-10460.

Kjaer M. Visual evoked potentials in normal subjects and patients with multiple

sclerosis. Acta Neurol Scand 1980;62:1-13.

Kugler C FA. Interrelations of age, sensory functions, and human brain signal

processing. J Gerontology 1999;54: B231-B238.

La Marche JA, Dobson WR, Cohn NB, Dustman RE. Amplitudes of visually evoked

potentials to patterned stimuli: age and sex comparisons. Electroencephalogr Clin

Neurophysiol 1986;65:81-85.

Lesevre N, Joseph JP. Modifications of the pattern-evoked potential (PEP) in relation

to the stimulated part of the visual field. Electroencephalogr Clin Neurophysiol

1979;47:183-203.

Lintl P, Braak H. Loss of intracortical myelinated fibers: a distinctive age-related

alteration in the human striate area. Acta Neuropathol (Berl) 1983;61:178-182.

Meier-Ruge W, Ulrich J, Bruhlmann M, Meier E. Age-related white matter atrophy in

the human brain. Ann N Y Acad Sci 1992;673:260-269.

Michael WF, Halliday AM. Differences between the occipital distribution of upper

and lower field pattern-evoked responses in man. Brain Res 1971;32:311-324.

Niedermeyer E, Lopes da Silva F. In: Electroencephalography: Basic principles,

clinical applications and related fields. Baltimore-Munich: Urban and Schwarzenberg,

1987:1-55.

Picton TW, Lins OG, Scherg M. The recording and analysis of event-related

potentials. In: Johnson Jr.R, Baron JC, editors. Handbook of neuropsychology Vol.10:

Amsterdam: Elsevier, 1995: 3-73.

Polich J. EEG and ERP assessment of normal aging. Electroencephalogr Clin

Neurophysiol 1997;104:244-256.

Polich J, Luckritz JY. EEG and ERP in young and elderly subjects. In: Karmos G,

Molnar M, Csepe V, Czigler I and Desmedt JS, editors Perspectives of Event-Related

Potentials Research (EEG Suppl. 44). Amsterdam: Elsevier, 1995:358-368.

Page 88: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

88

Porceddu ML, De Montis G, Pepitoni S, Toffano G, Biggio G. Failure of dark

adaptation to upregulate D-1 dopamine receptors in retina of senescent rats. Neurobiol

Aging 1990;11:105-109.

Pratt H, Bleich N, Berliner E. Short latency visual evoked potentials in man.

Electroencephalogr Clin Neurophysiol 1982;54:55-62.

Rahn E, Basar E. Prestimulus EEG-activity strongly influences the auditory evoked

vertex response: a new method for selective averaging. Int J Neurosci 1993;69:207-

220.

Schroder H, Giacobini E, Struble RG, Zilles K, Maelicke A, Luiten PGM, Strosberg

AD. Cellular distribution and expression of cortical acetylcholine receptors in aging

and Alzheimer's disease. Ann N Y Acad Sci 1991;640:189-192.

Semlitsch HV, Anderer P, Schuster P, Presslich O. A solution for reliable and valid

reduction of ocular artifacts, applied to the P300 ERP. Psychophysiology

1986;23:695-703.

Shaw NA. Changes in the cortical components of the visual evoked potential with age

in man. Aust J Exp Biol Med Sci 1984;62:771-778.

Shaw NA, Cant BR. Age-dependent changes in the amplitude of the pattern visual

evoked potential. Electroencephalogr Clin Neurophysiol 1981; 51:671-673.

Shaw NA, Cant BR. Age-dependent changes in the latency of the pattern visual

evoked potential. Electroencephalogr Clin Neurophysiol 1980;48:237-241.

Sidman RD, Major DJ, Ford MR, Ramsey GG, Schlichting C. Age-related features of

the resting pattern-reversal visual evoked response using the dipole localization

method and cortical imaging technique. J Neurosci Methods 1991;37:27-36.

Snyder EW, Dustman RE, Shearer DE. Pattern reversal evoked potential amplitudes:

life span changes. Electroencephalogr Clin Neurophysiol 1981;52:429-434.

Sokol S, Moskowitz A, Towle VL. Age-related changes in the latency of the visual

evoked potential: influence of check size. Electroencephalogr Clin Neurophysiol

1981;51:559-562.

Taylor M J. The role of event-related potentials in the study of normal and abnormal

cognitive development. In: Johnson R Jr., Baron JC, editors. Handbook of

neuropsychology Vol. 10. Amsterdam: Elsevier, 1995:187-211.

Tobimatsu S, Kurita-Tashima S, Nakayama-Hiromatsu M, Akazawa K, Kato M.

Age-related changes in pattern visual evoked potentials: differential effects of

luminance, contrast and check size. Electroencephalogr Clin Neurophysiol

1993;88:12-19.

Verma NP, Kooi KA. Gender factor in longer P100 latency of elderly persons.

Electroencephalogr Clin Neurophysiol 1984;59:361-365.

Page 89: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

89

Vrabec F. Senile changes in the ganglion cells of the human retina. Brit J Ophthal

1965;49:561-572.

Willis WD Jr. The cerebral cortex and higher functions of the nervous system. In:

Berne RM, Levy MN, editors. Physiology. St. Louis: Mosby, 1998: 249-266.

Woody CD. Characterization of an adaptive filter for the analysis of variable latency

neuroelectric signals. Med Biol Eng 1967;5:539-553.

Page 90: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

90

Page 91: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

91

Page 92: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

92

Page 93: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

93

Page 94: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

94

Page 95: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

95

Page 96: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

96

Page 97: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

97

Page 98: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

98

Page 99: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

99

Page 100: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

100

Page 101: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

101

Page 102: Technique for the Analysis of Evoked and Background EEG Activity applied to Young and Elderly Subjects - Doctoral Dissertation of Craig A. Goodman

102