Group Title: event-related potential "P3"
Title: The event-related potential "P3"
Full Citation
Permanent Link:
 Material Information
Title: The event-related potential "P3"
Physical Description: Book
Language: English
Creator: Sandridge, Sharon Ann, 1955-
Copyright Date: 1988
 Record Information
Bibliographic ID: UF00102758
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: ltuf - AFK0962
oclc - 19728671

Full Text









The formulation, implementation and completion of this

study, and thus my degree, have been made possible only

through the assistance and support of a number of dedicated

people. I would like to thank the members of my committee,

Drs. Kenneth Gerhardt (Cochair), Edward Hammond, Alice

Holmes, F. J Kemker, and Otto von Mering, for the expertise,

advice and encouragement they provided, not only over this

past year but over the entire tenure of my doctoral program.

My deepest thanks go to Drs. Sharon Lesner and Patricia

Kricos (Chair), my mentors and friends. Sharon encouraged

me to seek this degree, Pat guided me through it. I extend

my sincerest gratitude to both of them for their endless

hours of assistance and support.

Finally, what would acknowledgements be without

families. Although it may sound trite, I owe this degree to

my mother and father. I really could not have done it

without them. My mother was always there for me, and my

father's confidence and pride instilled an even greater

drive to complete this degree. I hope that my parents are as

proud to have me as a daughter as I am to have them as


And the most important thank you goes to my son,

Nicholas. I dedicate this degree to him because he has been

through it all with me; the endless hours spent in my office

so that Mommy could study and the countless nights sleeping

on the floor in the lab so that Mommy could write.

Nicholas has been the source of my strength and my sense of

being. He gave life the proper perspective and made me

"stop to smell the roses."

Thank you God for blessing me with Nicholas, my family

and friends.




ACKNOWLEDGEMENTS....................................... ii

ABSTRACT............. .................................. v

CHAPTER I INTRODUCTION.. ............................ 1

CHAPTER II REVIEW OF LITERATURE....................... 9

Electrophysiologic Potentials............... 9
The P3 Component........................... 16
Age-Related Changes of the Auditory
System.................................... 29
Behavioral Auditory Age-Related Changes.... 32
Summary........................ ......... ... 39

CHAPTER III MATERIALS AND METHODS....................... 42

Subjects. ................................. ..42
Audiometric Testing........................ 43
Electrophysiological Testing.............. 44
Procedures................................. 47

CHAPTER IV RESULTS.................................... 50

Subjects............. ......... ............ 50
Behavioral Measurements.................. 51
Summary of Behavioral Results.............. 60
Electrophysiological Results................ 62

CHAPTER V DISCUSSION................................ 74

The P3 Component in Quiet.................. 75
The P3 Component in Noise................. 77
The Mixed Group............. .......... .... 82
The P3 Component: Summary................ 86
Limitations of the Study................... 91

REFERENCES............. .............. .............. .... 94

BIOGRAPHICAL SKETCH ............ ...... ....... ......... 111

Abstract of Dissertation Presented to the Graduate
School of the University of Florida in Partial
Fulfillment of the Requirements for the Degree of
Doctor of Philosophy



Sharon Ann Sandridge

August, 1988

Chairman: Patricia B. Kricos, PhD
CoChairman: Kenneth J. Gerhardt, PhD
Major Department: Speech

The growth of the older population in America brings

with it a new set of problems. One such problem is the loss

of hearing sensitivity and perception. For many, the

ability to hear decreases with advancing age. For a

percentage of those adults, this decreased ability to hear

is accompanied by a decreased ability to understand speech,

a reduction which has been termed the "central auditory

aging affect."

Assessment of central auditory impairment involves the

administration of specific tests designed to evaluate the

integrity of the central auditory nervous system. One such

test used clinically is the Synthetic Sentence

Identification test. Another test, an event-related

potential test known as the P3 component, has been suggested

as a potential central auditory test. It was the purpose of

this study to investigate the use of the P3 as a measure of

central auditory function.

Thirty-one healthy and cognitively intact, male

veterans received a battery of behavioral audiometric

testing and P3 testing. The subjects were classified as

either centrally impaired (CI) or peripherally impaired (PI)

based on the results of the behavioral tests. The data from

the P3 testing, according to group classification, were

submitted for statistical analysis.

Within the limits imposed by the design of the study,

the following conclusions were drawn: (a) The latency of

the P3 component invoked in a quiet condition supported the

placement of the groups. The PI group latency was

consistent with previously reported age-related norms. The

CI group latency, however, was longer than the PI groups

indicating an increased processing time or extended stimulus

evaluation time. (b) The introduction of a medium band

noise served to increase the complexity of the stimulus as

evidenced by increased latencies for both groups. (c)

While both groups demonstrated an effect of the noise, the

PI group appeared to be more affected by the noise than the

CI group. (d) The results of this investigation support

the use of the P3 component, when invoked in quiet, as a

measure of central auditory function in an older, hearing-

impaired population.



Aging is a process that begins with conception and ends

with death. The early process of aging is characterized by

continual growth and development: an infant develops the

coordination to walk, a child develops the integration of

visual ability with cognitive ability to learn to read,

teenagers pass through puberty. On the other end of the

continuum the aging process is characterized by decline in

body function. The body and its functions undergo many

biological changes as they approach the end of life's

continuum. One such change involves the decreasing ability

to hear. Numerous investigators have attempted to document

the percentage of the older population with decreased

hearing sensitivity. For the older adult, defined by

convention as anyone 65 or older, the percentages vary from

a conservative figure of 30% (Fein, 1983) to a somewhat less

conservative figure of 83% (Moscicki, Elkins, Baum, and

McNamara, 1985). Sommers, Weidner, and McAleer (1982)

reported that 100% of the elderly population in their sample

of nursing home residents demonstrated a hearing impairment.

While the estimates of hearing impairments for the older

population vary depending on the criteria used for defining

hearing loss and the population sampled, there is little

doubt that hearing impairment in the older adult population

does exist.

There is also little debate over the issue that as age

increases there is a concomitant increase in incidence rate.

According to the National Center for Health Statistics

(1981, 1982) the rate of hearing impairment increased from

24% for 65- to 74-year-olds to 39% for persons over 75-years

old. Dalzell and Puccia (1985) noted that in their young-

old (65- to 69-year-old) group, there was an incidence rate

of 56% for hearing loss; this increased to 89% for the

oldest-old (85 and older) group.

The prevalence for decreased hearing ability,

therefore, increases as a function of age. Yet, the hearing

problems experienced by the elderly are more complex than

just a loss of sensitivity. In addition to the decline in

hearing sensitivity, the elderly experience more difficulty

comprehending speech. Their ability to understand speech,

especially in situations where speech is degraded, is

reduced. This decreased ability to understand, especially

in difficult listening situations, has been attributed to an

impairment in the central auditory system and has been

referred to as the "central auditory aging effect" (Jerger &

Hayes, 1977). The older hearing impaired individual may,

thereby, demonstrate a central auditory nervous system

disorder in addition to the impairment of the peripheral

hearing mechanisms. Hodgson (1972, p. 313) defined a


central auditory nervous system disorder as a deficit in the

"process of formal integration in the relays situated at

different stages along the auditory pathway." It is not a

global deficit of the central nervous system; it is a

dysfunction localized to the auditory system. Generally, a

central auditory aging effect occurs in the presence of an

intact cognitive system.

Jerger (1973) has specifically characterized the

hearing deficit associated with aging as involving both the

peripheral auditory system (evidenced by factor 1 below) and

the central auditory system (evidenced by factors 2, 3, and

4) and is indicated by the following factors:

1. Hearing loss with elevation of thresholds for pure

tones and speech.

2. Reduced ability to understand speech.

3. Reduced ability to transmit complex speech signals.

4. Retrocochlear testing non-suggestive of eighth

nerve pathology.

Hearing loss in the older population, therefore, involves

both the decreased ability to "hear" as well as the

decreased ability to "understand."

This reduced ability to understand speech and complex

speech signals as a function of aging has been widely

investigated using various methodologies. For example,

speech in the presence of a competing noise, filtering the

speech signal and compressing the speech signal have been

used to demonstrate this aging effect. In a number of

articles by Jerger and Hayes (Jerger, 1973; Jerger & Hayes,

1973; Hayes & Jerger, 1979a; Hayes & Jerger, 1979b), the

central auditory aging effect has been demonstrated by

comparing intensity functions obtained for phonetically

balanced monosyllabic words (PB) with functions obtained for

synthetic sentences presented with a competing story (SSI).

Results from their series of studies indicated that the

older adult generally performs poorer than the younger adult

on both the PB words and the SSI. Results also indicated

that some, but not all, of the older adult listeners

demonstrated a more substantial decline on the SSI function

than would be expected or anticipated from the audiogram or

PB scores. Jerger and Hayes concluded that the older

adults who performed more poorly on the SSI relative to the

PB words demonstrated a more central auditory deficit. This

discrepancy score between the maximum score obtained on the

PB words and the maximum score obtained on the SSI has

become an acceptable clinical tool for classifying an older

individual as having primarily a peripheral deficit or a

combined peripheral and central impairment. Jerger and

Hayes suggested that if the performance on the SSI was

poorer by 20% or more than the PB words, the individual

exhibited central as well as peripheral involvement.

The Synthetic Sentence Identification (SSI) Test is one

of the most widely used measures of central auditory

processing (Arnst, 1986), yet it is one among many tests

designed for this function. Recently, more objective

electrophysiologic tests have been investigated as measures

of central auditory function, including the event-related

potential known as P3 or P300.

Squires and Hecox (1983) reported several case studies

supporting the use of P3 as a measure for central auditory

function. For example, a 27-year-old woman was referred for

audiological testing due to complaints of decreased speech

comprehension in the presence of background noise (a typical

complaint of the older adult). A standard audiometric

evaluation was administered; results were normal. Brainstem

auditory response, middle latency response and long latency

(P3) response testing were also normal. Unsatisfied with

the results (obtained in quiet) and their suggestion of

normalcy, coupled with the patient's complaints of decreased

ability to understand speech in the presence of noise,

electrophysiological testing was repeated using a minimal

amount of background noise (30 dB signal-to-noise ratio).

The brainstem and middle latency responses were unaltered by

the noise. The results of the long latency potentials were,

however, strikingly altered. The P3 response was eliminated

in the presence of noise. Additional audiometric testing in

the presence of noise supported the electrophysiological

results. This patient was diagnosed as having a cortical

pathology and not a psychiatric diagnosis, at one time a


The P3 component is a positive going component occurring

around 300 ms post stimulus onset for young adults. It is

an endogenous component. That is, the potential referred to

as P3 only occurs when the stimulus information is being

"actively processed." It is independent of stimulus

modality and stimulus parameters. It is instead, dependent

on psychological variables. Therefore, stimulus parameters

such as rise/fall time and intensity are not variables of

the endogenous response and will not affect the response.

On the other hand, the degree of attending, motivation, and

task relevancy are variables of the endogenous response and

will affect the response.

As Squires and Hecox (1983) suggested, the P3 component,

specifically the latency of the P3 component, could be used

for detecting certain deficits in auditory comprehension.

The latency of the P3 component has been strongly suggested

to be an index of stimulus evaluation time and very subtle

changes in the amount of time required to process

information are correlated with variations in the latency of

the P3 component. Sometime prior to the occurrence of the

P3 component, the stimulus is encoded, identified and

evaluated. By changing the complexity of the stimuli or by

increasing the difficultly in which the stimuli are

discriminated or detected, a concomitant change in the

latency of P3 occurs; as complexity increases so does the

latency. Polich, Howard, and Starr (1985a) changed the

complexity of their task by introducing a background or

masking noise with the test stimuli. The noise served to

degrade or interfere with the primary stimulus resulting in

a need for increased processing time and hence, a longer P3

component latency.

For many older adults, the chief complaint regarding

their hearing is the inability to understand speech in the

presence of background competition. The SSI attempts to

simulate that condition by presenting the message in the

presence of the competing story. For the P3, the

introduction of a background noise should theoretically serve

the same purpose; in both situations the noise (speech or

bandpass noise) interferes with the processing of the

stimuli. While the P3 component has been widely

investigated by other disciplines, it has not been

investigated extensively in the field of audiology and has

not been employed to investigate the central auditory aging

effect in the older population. It was, therefore, the

purpose of this study to investigate the use of the P3

latency measure evoked in both quiet and noise conditions as

a measure of central auditory dysfunction in the older

hearing-impaired population. The following questions were


1. Will subjects who have been classified as

peripherally impaired based on the PB-SSI MAX criterion show

latency values for the P3 component that are consistent with

the previous reported age-related norms for 65- to 75-year-

olds for the quiet condition?

2. Will subjects who have been classified as centrally

impaired based on the PB-SSI MAX criterion show latency

values for the P3 component, in quiet, that are

significantly longer than the latencies of the PI subjects,

thus supporting the hypothesis that centrally impaired

individuals need more time for processing?

3. Will the use of noise be sufficient to increase the

difficulty of the discrimination as evidence by

significantly longer latencies for the P3 component evoked

in noise compared to the quiet condition?


A review of the literature relevant to this

investigation of the P3 component as a central auditory test

for the older adult will be presented. Specifically,

research dealing with the electrophysiologic potential known

as "P3", characteristics of the various peripheral and

central changes that occur as a function of aging, and their

effects on the auditory system will be examined.

Electrophysiologic Potentials

Over 100 years ago, Richard Caton (1875) first noted

the invariable presence of electrical activity in the brain

and reported the existence of both spontaneous brain rhythms

and evoked potentials. Spontaneous brain rhythms are the

random fluctuations of electrical activity know as electro-

encephalograms (EEG). Evoked potentials are not

"spontaneous" but rather time-locked activity resulting from

stimulation of the sense organs or an afferent pathway. The

strength of the evoked potentials at the human scalp,

however, is so small (0.1 to 10uv) compared with the ongoing

electroencephagraphic activity (10 to over 100 uv) that

until the development of signal averaging, evoked potentials

could only be recorded in cases of brain surgery where


direct scalp electrode placement was possible. With the

development of the technique for summing and averaging the

brain's electrical responses to repeated stimuli (Dawson,

1947), evoked potential testing was made possible using far

field electrode placement. Further advances in computer

technology have facilitated the use of evoked potential

testing. Today, evoked potential testing is commonly used

to evaluate the central nervous system (e.g. Fria, 1980;

Chiappa & Ropper, 1982).

Evoked potentials can be elicited from any discrete

auditory, visual, or somatosensory stimuli and consist of a

series of electrical components. For example, in response

to an auditory stimulus, 16 oscillations can be reliably

recorded from the scalp (Hillyard, Picton, & Regan, 1978)

with the earliest appearing within a few milliseconds of

stimulus onset and extending for several hundred milli-

seconds thereafter (Figure 1).

The earliest appearing auditory potentials are commonly

referred to as "the early potentials" or "brainstem

potentials." Within the first 10 ms, a series of six or

seven discrete waves, labeled I VII (Jewett & Williston,

1971), can be recorded in response to an auditory click or

tone pip. These potentials are of submicrovolt size and are

believed to represent the neuronal activity of the auditory

pathway from the eighth nerve to the rostral brainstem

(Jewett, Romano, & Williston, 1970; Buchwald &

Pa Pb

I \


U'V v V v V
Nb N1

I N2



Figure 1. An example of the electrophysiological response
to an auditory stimulus. The early components
occur within the first 10 ms, the middle
components occur between 10 and 100 ms, and the
late components occur 100 ms poststimulus onset.
Note: The amplitude is not drawn to scale.


Huang, 1975; Starr & Achor, 1978). Early evoked potentials

are sensitive to stimulus parameters, particularly spectrum,

intensity, rise time, and repetition rate (Hecox, Squires, &

Galambos, 1976; Hyde, Stephens, & Thorton, 1976; Pratt &

Sohmer, 1976); however, they do not change appreciably with

changes of mental state, i.e. sleep, (Amadeo & Shagass,

1973) or attention (Picton & Hillyard, 1974; Woods &

Hillyard, 1978). The auditory brainstem potentials have

been investigated extensively and provide a sensitive and

objective means of assessing hearing deficiencies due to

cochlear pathology or neurological conditions involving the

auditory pathway within the brainstem.

Following the early components, a series of negative

and positive deflections ranging from approximately 10 to

100 ms occur, referred to as the "middle latency" components

(Goldstein & Rodman, 1967). The origins for the middle

components are unclear. Intracranial recordings from humans

suggest that the earliest middle latency potentials

originate at the midbrain level (Hashimoto, 1982); whereas

the later middle components may have a thalamic and/or

cortical origin (Kaga, Hink, Schinoda, & Suzuki, 1980;

Buchwald, Hinman, Norman, Huang, & Brown, 1981). The

specific generators remain to be established.

The brainstem and middle components are also referred

to as "exogenous" or "stimulus-bound" potentials. The

characteristics of exogenous potentials, mainly the latency


and amplitude of the potential, are primarily determined by

the physical characteristics of the stimulus and are

relatively insensitive to changes in psychological

variables. Exogenous potentials are "evoked" by events that

are extrinsic to the nervous system. They represent the

summation of synchronous activation of neuronal populations

by an external stimulus event.

The third class of electrical potentials seem,

however, to be largely independent of the physical

parameters of the triggering stimuli and to reflect instead

the individual's attentiveness, expectancies, and

information processing strategies. Such potentials have

been termed endogenouss" (Sutton, Braren, Zubin, & John,

1965). Endogenous potentials depend very critically upon

the context within which a stimulus is presented and the

psychological state of the individual. They are related to

"internally" generated events, not to the "externally"

generated event. While endogenous potentials are "invoked"

by an external event (be it visual, auditory, or

somatosensory), the characteristics of exogenous potentials,

namely amplitude and latency, are determined by the

particular cognitive processes activated by the stimulus

rather than by its modality or physical properties.

Generally, the series of components occurring after 100 ms

poststimulus onset are considered to be endogenous and

broadly categorized as "late" potentials.

The event-related potentials recorded from the human

scalp are complex. To facilitate understanding of this

complex waveform, individual components can be identified.

There is, however, no universally accepted technique for

identifying and validating the separate components of a

recorded waveform.

An initial approach is to identify and label the peaks

and troughs of the waveform. Peaks are identified on the

basis of maximum amplitude inflection in both spatial and

temporal dimensions (Lehmann, 1977). These peaks and

troughs are typically labeled with a "P" indicating a

positive source or an "N" indicating a negative sink. The

letters are then coupled with either an approximate latency

value (e.g., P300; suggesting that a positive peak occurs

around 300 ms poststimulus onset) or a number indicating the

position in the waveform (e.g., P3; indicating it is the

third positive peak of the waveform). The advantage of

identifying and labeling the peaks and troughs is one of

convenience. It is difficult, however, to determine the

functional significance of the identified peak. It is not

known whether a particular scalp-recorded peak represents

one particular cerebral process or whether a particular peak

results from the superimposition of multiple underlying


In order to resolve this question, the underlying

physiological generators for the scalp-recorded potentials


need to be determined and there are basically two approaches

used. One approach deals with the delineation of

"functional neuroanatomy." This involves correlating

intracranial recordings from animals and humans with scalp-

recorded potentials. While the advantages of this procedure

are numerous, it has not yet been possible to determine with

any certainty the generator sources for most human scalp-

recorded potentials (Picton & Stuss, 1980).

The more commonly used approach involves the manipu-

lation of experimental variables. Any portion of a recorded

waveform that can be independently manipulated by experi-

mental variables is considered to be a distinct component.

Instrumental variables such as the type of stimulus (click,

tone), stimulus intensity, repetition rate, and electrode

placement, may be varied and will have more dramatic results

with exogenous potentials. For endogenous potentials, the

psychological variables need to be manipulated. For

example, attention can be active or passive, the stimulus

can have relevance or no relevance, the gender and age of

the subjects can be manipulated, and the memory load can be

increased or decreased. The advantages of variable

manipulation may be the objectivity and the ability to

evaluate components that overlap in time or space. On the

other hand, waveform components may be described that have

only a loose connection or relationship to the possible

underlying generator source. Nevertheless, the manipulation

of variables seems to be the most efficient means of

determining the components' significance (Picton & Stuss,


The P3 Component

The most widely investigated of the endogenouss

components" is the P300 or P3 component. The P3 component

is a positive peak that occurs somewhere from 220 ms to

900 ms poststimulus (Kutas & Hillyard, 1984). Sutton

et al. (1965) reported that a component of the human

auditory evoked potential exists that is not dependent on

the physical parameters of the eliciting stimuli, but

rather, is elicited by task-relevant, "surprising" stimuli

and appears to reflect active cognitive processing of the


The P3 component is generally and most easily invoked

using an "oddball" paradigm. That is, the subject detects

infrequent "target" (rare) stimuli occurring randomly within

a series of non-target (frequent) stimuli and keeps a mental

count or responds motorically to the target or rare stimuli.

The P3 component accompanies the detection of the targets.

Since the P3 component is an endogenous potential, it is

modality nonspecific and thus can be elicited with visual

(i.e. Kok & Looren de Jong, 1980), auditory (i.e. Polich,

Howard, & Starr, 1985a) or somatosensory stimuli (i.e. Wood,

Allison, Goff, Williamson, & Spencer, 1980). In addition,

the P3 component has been elicited by pitch differences

(Ritter, Simson, & Vaughan, 1972), intensity differences

(Papanicolaou, Loring, Raz, & Eisenberg, 1985; Roth, Doyce,

Pfefferbaum, & Kopell, 1980; Snyder & Hillyard, 1976),

probability differences (Polich, 1986b), difference of

memory demands (Donald & Young, 1982; Johnson, Pfefferbaum,

& Kopell, 1985; Megela & Teyler, 1979), and semantic aspects

of speech (Bentin, McCarthy, & Wood, 1985; Harbin, Marsh, &

Harvey, 1984; Kutas, McCarthy, & Donchin, 1977; Novick,

Lovrich, & Vaughan, 1985). To emphasize further the endo-

genous aspect of this component, P3 can even be elicited by

stimuli that are expected and anticipated but which do not

occur or are omitted (Klinke, Fruhstorfer, & Finkenzeller,

1968; Picton, Hillyard, & Galambos, 1974; Ruchkin & Sutton,

1979). An example of a recorded P3 component is shown in

Figure 2. Note that the P3 component is seen primarily in

the average of the rare stimuli as a large positive

deflection following the second positive peak.

Waveforms of electrophysiological potentials are

commonly described in terms of amplitude and latency.

Recall that the amplitude and latency of endogenous poten-

tials are determined by the particular cognitive processes

activated rather than the physical parameters of the

eliciting stimulus. Specifically, the amplitude of the P3

component has been shown to be a function of the confidence



0 375

Figure 2.

A recorded response of the P3 component
with replication is shown. Note that the
top three tracings are the averaged
response to the frequent (common) tone and
do not illustrate the P3 component. The
bottom three tracings are the averaged
response to the rare tones and exhibit the
P3 component. The tracings labeled "Fz"
are recorded from the frontal electrode
site, the "Cz" are recorded from the
vertex electrode site, and the "Pz"
tracings are from the parietal electrode




< Pz

750 ms


of detection (Campbell, Courchesne, Picton, & Squires, 1979;

Squires, Hillyard, Lindsay, 1973; Squires, Squires, &

Hillyard, 1975) and the subjective probability of the target

stimuli (Duncan-Johnson & Donchin, 1977; Squires, Wickens,

Squires, & Donchin, 1976; Tueting, Sutton, & Zubin, 1970).

The more infrequent and unexpected the target, the larger is

the amplitude (Duncan-Johnson & Donchin, 1977; Sutton et

al., 1965). In fact, in certain situations, the voltage is

of such magnitude (20-63 uv) that signal enhancement

techniques are not necessary and individual waveforms may be

studied (Cooper, 1981). It is, however, important to note

that it is the subjective probability of the stimulus

occurrence rather than the objective probability that is the

determination of P3 amplitude. That is, the subject's

expectancy concerning an event will determine the amplitude

of the P3 component (Johnson & Donchin, 1980; Squires et

al., 1976). Duncan-Johnson and Donchin (1980) demonstrated

that different stimuli presented with equal expectancy yield

the same P3 component, yet the same stimuli presented with

different expectancies yield different P3s. Therefore, it

is the perception of probability that determines P3

amplitude rather than the objective physical structure of

the stimulus series.

While amplitude of P3 is affected by subjective prob-

ability, task relevance, and confidence of detection, it has

not yet been clearly demonstrated that stimulus complexity

influences P3 amplitude (Johnson, 1984). The influence of

stimulus complexity on the latency of the P3 component,

however, has been well documented.

The second parameter of the P3 waveform is peak

latency. As the stimulus becomes more difficult to encode

or becomes more perceptually complex, the latency of the P3

component increases (Ford, Mohs, Pfefferbaum & Kopell, 1980;

Kok & Looren de Jong, 1980; McCallum, 1980; Squires,

Donchin, Squires, & Grossberg, 1977). For example, McCallum

(1980) increased the complexity of the task involving

successive 1 sec pseudo-random presentations of a square,

cross, or circle. Five task conditions of progressing

difficulty were employed: Task "a" required no response to

the stimuli while Task "e" required a motoric response only

when a circle was followed by a cross, a cross followed by a

square, or a square followed by a circle. Peak latencies

increased markedly from 325 ms for the easiest tasks to 390

ms for the most difficult task.

A number of theories have been proposed to account for

the latency of the P3 component. The literature presented

seems to consistently support that the latency of the P3

component is determined by stimulus-evaluation time (Duncan-

Johnson, 1981; Duncan-Johnson & Donchin, 1982; Johnson and

Donchin, 1985; Kutas, McCarthy, & Donchin, 1977;

Pfefferbaum, Ford, Johnson, Wenegrat & Kopell, 1983). The

theory proposed by Donchin and his co-workers asserts that

certain stimulus evaluation activities must be completed

prior to the process that is reflected by P3. It does not

mean that P3 latency is an actual measure of stimulus

evaluation time. It simply means that if a stimulus is

going to be evaluated, it is evaluated before the P3

component occurs. Thereby, if stimulus evaluation is

manipulated, P3 latency will also be manipulated since it is

dependent upon that process.

Research has, in fact, demonstrated that by increasing

the stimulus evaluation process, P3 latency is increased.

One method to increase stimulus evaluation is to increase

the difficulty of the task. For example, Ritter et al.

(1972) increased the task difficulty by narrowing the

frequency difference between the target and nontarget tones.

As a result the latency of P3 was prolonged. Squires et al.

(1977) studied the effects of task difficulty in both

auditory and visual modalities and found a marked effect on

discriminability of P3 latency. In the auditory task, the

latency of P3 to an 1100 Hz tone occurred 60 ms later when

paired with a 1060 Hz tone than when paired with a less

similar 1000 Hz tone. Duncan-Johnson and Kopell (1981)

demonstrated the same effect with visual stimuli. They

presented subjects with a series of slides that varied in

hue discriminability. As the hues became less discernable,

P3 latency increased, an average of 80 ms.

Similarly, P3 latency is increased when the target

stimulus is embedded in a noisy background. Polich, Howard,

and Starr (1985a) investigated the effects of increasing the

frequency differences between the target and nontarget tones

and the effect of the presence of white masking noise. The

rare tone was held constant at 1000 Hz whereas the frequent

tones were either 1500, 2000, or 4000 Hz. The tones were

presented in quiet and in the presence of a 60 dB nHL white

noise. Results supported previous findings that as the

difference in frequencies decreased, therefore becoming more

difficult to discriminate, the latencies became longer. In

addition, when the masking noise was present, the same

effect was seen; however, the latencies were more prolonged

suggesting that the time needed to evaluate the tone in the

masking noise was longer. Ford et al. (1980) manipulated

the quality of the stimulus as well as the size of the

memory set and found similar results. As the memory set

increased from one to four, P3 latency increased. As the

quality of the stimulus was degraded, the latency of the P3

component also increased. Comparison of the slopes and

intercepts of these functions suggested that the encoding of

the degraded stimulus was completed before it was compared

to the items in memory. Kok and Looren de Jong (1980)

presented slightly different results when they investigated

the effects of degraded visual stimuli on the latency of the

P3 component. They failed to demonstrate any significant

increase in P3 latency when the visual stimulus was


Additional support is provided for the stimulus eval-

uation hypothesis by a number of studies that investigated

the relationship between the P3 component and reaction time

(RT). While the majority of investigators have reported

positive correlations between RT and P3 latency (i.e.

Bostock & Jarvis, 1970; Ritter et al., 1972; Rohrbaugh,

Donchin, & Eriksen, 1974; Roth, Ford & Kopell, 1978), the

results of a number of well-controlled studies have

indicated a dissociation between P3 latency and reaction

time (Karlin & Martz, 1973; Karlin, Martz, Brauth, &

Mordkoff, 1971; Wilkinson & Morlock, 1967). McCarthy and

Donchin (1981) provided further support for the dissociation

between RT and P3 latency. Subjects in their study were

required to indicate the identity of the target word (right,

left) by pressing the appropriate button. Both the stimulus

and response task were made more difficult. Stimulus

discriminability was reduced by embedding the target word in

a matrix of letters. The stimulus-response task was

increased in difficulty by cueing the subject as to which

thumb to use to respond. That is, the subjects were cued by

the words "same" or "opposite" and were to respond with the

right thumb if the target word was "right" and the cue word

was "same." Results indicated that RT was influenced by

both variables (stimulus discriminability and stimulus

response compatibility) whereas P3 latency was only

influenced by the stimulus discriminability. It appears

then, that the latency of the P3 component is a pure measure

of stimulus evaluation time and is unencumbered by response

selection and execution time.

The Origin of the P3 Component

The neural origin of the P3 component is not known.

Numerous investigators, employing such techniques as scalp

topography, depth electrode recordings, and animal models,

have proposed various origins. From their investigations,

four hypotheses have evolved to account for the origin of

P3. Several researchers suggest that the generator site is

the hippocampus or surrounding areas (Begleiter, 1979;

Halgren, Squires, Wilson, Rohrbaugh, Babb, & Crandall, 1980;

O'Connor & Starr, 1985). A second theory involves a purely

cortical origin (Goff, Allison, & Vaughan, 1978; Simson,

Vaughan, & Ritter, 1976; Simson et al., 1977a; Simson et

al., 1977b). A third hypothesis is the reticulo-thalamo-

cortical activating system which involves the mesencephalic

reticular formation, medial thalamus, and prefrontal cortex

(Desmedt, 1981; Desmedt & Debecker, 1979; Yingling &

Hosobuchi, 1984). The fourth theory is one of multiplicity

(Wood and Wolpaw, 1982); that is, there may be several

neural generator sites for the P3 component.

The early theories postulated that since the P3

component has a broad central-parietal scalp distribution,

then it is generated by a diffuse cortical source (Goff et

al., 1978; Simson et al., 1976; Simson et al., 1977a; Simson

et al., 1977b). Wood et al. (1980) later cast doubt upon a

purely cortical origin after recording, using depth

electrodes, from 12 patients with intractable epilepsy.

From their results, they suggested that P3 activity is

likely to have a subcortical generator site.

A number of investigators have proposed the hippocampal

area as the neural site. Halgren et al. (1980) recorded

from an epileptic patient and noted polarity reversal

between the hippocampus and parahippocampal gyrus. Squires,

Halgren, Wilson, and Crandall (1983) further supported the

limbic system as origin from additional depth electrode

recordings. There was, however, considerable variability in

the recorded latencies. Yingling and Hosobuchi (1984) noted

that the latency values in the Halgren et al. study were

longer than the scalp P3 and possibly could be related to

the slow wave and not the P3 itself. O'Connor and Starr

(1985) suggested that the latency variability may be

indicative of multiple generator sites. In addition to

proposing a hippocampal site, O'Connor and Starr (1985),

recording from depth electrodes inserted into cats, noted

another polarity reversal and steep voltage gradient as the

electrode penetrated the posterior suprasylvian gyrus,

suggesting at least two sites of origin, at least in the


Also using depth electrodes, Yingling and Hosobuchi

(1984) recorded polarity reversal and latency values for P3

that were similar to the scalp recordings; however, the

electrodes were placed at levels above the hippocampus

suggesting a nonhippocampal site. From their findings, they

proposed a site that was more medially located such as the

thalamus or other surrounding structures rather than a

hippocampal site.

The findings presented are inconclusive as to the

generator site. O'Connor and Starr (1985) suggested that

the large variance in recording data may be suggestive of

multiple generator sites. Harrison, Buchwald and Kaga

(1986) demonstrated that the primary auditory cortex is not

necessary for generation of auditory P3-like response in the

cat yet suggested that the association cortex or more

subcortical systems may be involved. The P3 potential is an

index of a multiple process invoked by any sensory system.

It seems, therefore, that the generation of P3 could be from

multiple sites or structures.

Age and the P3 Component

With age, there is a decline in the number of function-

ing neuronal units and a decrease in myelin (Shock, 1983).

As a result, there appears to be an decrease in the neural

conduction velocity. That is, with age there appears to be

a general slowing of the speed at which neural impulses

travel. This slowing may result in a longer decision

process. Rabbitt (1965) suggested that, in fact, the

elderly are slower in deciding whether information is

relevant. Investigations of P3 latency as a function of age

support these findings (Brown, Marsh, & LaRue, 1983; Ford,

Hink, Hopkins, Roth, Pfefferbaum, & Kopell, 1979; Ford,

Roth, Mohs, Hopkins, & Kopell, 1979; Goodin, Squires,

Henderson & Starr, 1978; Pfefferbaum, Ford, Roth, & Kopell,

1980a; Pfefferbaum et al., 1980b; Pfefferbaum, Wenegrat,

Ford, Roth, & Kopell, 1984; Picton, Stuss, Champagne, &

Nelson, 1984; Syndulko, Hansch, Cohen, Pearce, Goldberg,

Monton, Tourtellotte, & Potvin, 1982)

Polich, Howard, and Starr (1985a) investigated the

changes in latency of the P3 component across age using the

classic oddball paradigm. One hundred and four subjects

were tested, ranging in age from 5 to 86. The latency was

found to decrease from the 0- to 9-year-old group (mean

latency was 355 ms) to the 20- to 29-year-old group (mean

latency of 287) and then gradually increase with the

advancement of age. For the oldest age group tested (70- to

90-year-olds, n = 8) a mean latency of 387 ms was obtained.

Results of this study were consistent with previous

investigations (Brown et al., 1983; Pfefferbaum et al.,

1984; Picton et al., 1984) showing a 1.3 ms per year

increase in latency. This value was slightly less than the

1.8 ms per year increase proposed by Goodin et al. (1978)


and slightly larger than the increase per year rate reported

by Squires, Chippendale, Wrege, Goodin and Starr (1980).

Polich et al. (1985a) also noted variability in the

latency measure. It appears that variability is minimal

(Polich, 1986a) and remains stable up to age 50 years and

then increases with age (Howard & Polich, 1985; Pfefferbaum

et al., 1984). Since clinical definition of abnormality is

often based on P3 latency deviation from the mean, it is

important to note that "normal" variation is likely to


In addition, the increase in latency with age appears

to be related to task difficulty. Squires et al. (1980)

investigated interaction between age and task difficulty in

44 subjects (age 8 to 82). Primary results indicated that

P3 latency increased with age in both the easy and the

difficult task. The magnitudes, however, differed. The

slope of the function for the easy task was .79 ms per year

while the slope for the difficult task was 1.49 ms per year.

The time needed for stimulus evaluation, as indexed by

the P3 component latency, thus increases as a function of

aging. In addition, it appears that the perceptual

processes associated with more difficult discriminability

are more susceptible to age than the easier tasks.

Age-Related Changes of the Auditory System

Peripheral Changes

Numerous subtle changes occur in the outer, middle and

inner ear as a function of aging. The changes of the outer

and middle ear, such as loss of elasticity of the external

auditory meatus, tympanic membrane and the middle ear

muscles and ligaments, have little to no effect on hearing

sensitivity. The age-related changes of the inner ear,

however, are somewhat more complex and do directly affect

hearing sensitivity.

Although the research is equivocal, there are several

consistent findings of changes in the inner ear that are

presumably the result of the aging process. The most

consistent finding is the loss of inner and outer hair cells

and supporting structures from the basal end (area for high

frequency sensitivity) of the first turn (Bredberg, 1968;

Hawkins & Johnsson, 1985; Johnsson & Hawkins, 1972a, 1972b,

1972c, 1979). Secondary to the loss of hair cells is the

corresponding degeneration of nerve fibers (Crowe, Guild &

Polvogt, 1934).

Schuknecht (1955, 1964), based on postmortem histo-

pathological studies and audiometric data, proposed that

age-associated changes of the cochlea could be classified

into one of four distinct categories. Schuknecht revised

his original categories in 1974. They are (1) sensory

presbycusis, characterized by hair cell loss, neural


degeneration, and atrophy of the organ of Corti; (2) strial

presbycusis, characterized by atrophy of the stria

vascularis and decreased supply of nutrition to the cochlea;

(3) neural presbycusis, characterized by primary

degeneration of cochlear neurons rather than the sensory

elements; and (4) cochlear conductive, characterized by

changes in the mechanical properties of the basilar membrane

(See Maurer & Rupp, 1979 or Schuknecht, 1975, for further

detail). While this classification system has been widely

accepted, it is not without controversy (Hawkins & Johnsson,


Central Auditory System

The central auditory nervous system begins as the

eighth cranial nerve enters the brainstem at the level of

the junction of the pons and the medulla and proceeds to the

auditory cortex in the temporal lobe. The central auditory

system functions to receive temporal, frequency, and

intensity information from the eighth nerve and encodes it

into messages that are perceived as meaningful.

Although research has been conducted on anatomical and

physiological changes associated with aging in the central

auditory system, the number of studies is few and the

results are equivocal. The cochlear nucleus, the most

widely investigated of the central auditory nuclei, has been

reported to show an uniform atrophy and degeneration of the

ganglion cells in the ventral cochlear nucleus (Feldman &

Vaughan, 1979; Kirikae, Sato, and Shitara, 1964). Limited

investigations of the superior olivary complex have

demonstrated the presence of degenerative cells (Hansen &

Reske-Nielsen, 1965) interspersed among normal cells and the

accumulation of pigment in some of these cells (Kirikae et

al., 1964) in the elderly. Ferraro and Minckler (1977a)

investigated the aging process of the lateral lemniscus and

noted that in the oldest subjects, there was evidence of a

decrease in the total number of axons. Ferraro and Minckler

(1977b) also investigated age-related changes in the

inferior colliculus and noted essentially no differences

between the age groups with the exception of a slightly

reduced fiber density. However, Hansen and Reske-Nielsen

(1965) reported reduced neuronal density and neuronal

degeneration in the inferior colliculus. Consistent with

the other brainstem structures, reduction of the number of

cells, degeneration of the cells and the presence of

increased pigment within other cells of the medial

geniculate body have been noted (Hansen & Reske-Nielsen,

1965; Kirikae et al., 1964).

While the age-related changes to the brainstem and

thalamic areas are equivocal, there is a consistent finding

of age-related changes in the cortex. There is a general

loss of neurons and thinning of the cortex (Brody, 1955;

Dublin, 1976; Hansen & Reske-Nielsen, 1965; Scheibel,

Lindsay, Tomiyasu, & Scheibel, 1975). Brody (1955) reported

that the oldest subjects had as much as 50% reduction in

neurons per unit area as the younger subjects.

Brody (1955, 1985) also noted similar cell number

reduction in the frontal lobe, occipital lobe, and the

precentral gyrus (interestingly, there was no significant

change in the posterior central gyrus), suggesting a global

reduction of neuronal cells in the cortex as a function of


Essentially, there appears to be a decreased number of

neurons accompanied by degeneration of some cells and the

accumulation of pigment in other cells. The specific types

of cells susceptible to damage remains unidentifiable at

this time. Yet intuitively, it would seem that the loss of

healthy functioning cells, from degeneration, atrophy or

pigmentation, will result in decreased ability to encode and

synthesize the complex impulses, resulting in degradation of

the incoming auditory message.

Behavioral Auditory Age-related Changes

Pure Tone Sensitivity

A hearing loss defined as presbycusis is most commonly

characterized by a gradually sloping, high-frequency loss

that is typically bilateral and symmetrical in nature

(Corso, 1963; Glorig & Davis, 1961; Glorig & Nixon, 1960;

Hinchcliffe, 1959; Spoor, 1967). It has been well

documented that the loss begins gradually and accelerates

with advancing age (Corso, 1963; Glorig & Davis, 1961;

Glorig & Nixon, 1960; Hinchcliffe, 1959; Moscicki et al.,

1985; Spoor, 1967). Dalzell and Puccia (1985) reported that

the mean hearing loss (average of 1000, 2000, and 3000 Hz)

was 29 dB HL for the 65-year-old group and increased to 64

dB HL for the 95-year-old; slightly more than 1 dB per year

increase in loss of sensitivity. Further analysis of the

increase indicated a .8 dB increase for the younger-old

group while the oldest-old group's hearing loss increased an

average of 1.5 dB per year.

Speech Discrimination

In addition to the decrease in pure tone sensitivity,

it is generally assumed that the ability to understand

speech diminishes as age increases. Gaeth (1948) was the

first to observe this phenomenon noting a disproportionate

loss in intelligibility of "common" words in hearing-

impaired subjects over the age of 60. Pestalozza and Shore

(1955) reported that subjects age 60 years and older tended

to score 20 to 40% poorer on speech discrimination tests

than younger subjects (age 50 years and younger). Luterman,

Welsh, and Melrose (1966), while controlling for hearing

loss, also reported a decrease in speech discrimination

ability as a function of age utilizing a young group (ages

20-38), and an older group (age 79-87).

Jerger (1973) studied the effects of aging on speech

discrimination using a large population (4095 ears from 2162

subjects) and found a systematic decrease in the maximum

score for phonetically balanced monosyllabic words (NDRC

word lists) with increase in age. Bess and Townsend (1977),

also employing a relatively large sample (742 ears), found a

significant difference in speech discrimination scores as a

function of age. They further noted that as the degree of

hearing loss increased there was an associated dramatic

decrease in speech discrimination ability with age.

While these studies and others (i.e. Blumenfeld,

Bergman, & Millner, 1969; Farrimond, 1961; Feldman & Reger,

1967) provide support for decreased speech discrimination

they are not accepted unequivocally. Surr (1977) carefully

controlled and matched audiograms and found no significant

differences between ages (30-90 years old) for mild high-

frequency hearing losses. Kasden (1970) compared audiograms

of 20- to 40-year-old and 60- to 69-year-old. Using the CID

W-22 words presented from 10 to 50 dB SL, Kasden also found

no age effect on speech discrimination scores.

Although there is a degree of uncertainty as to the

effect of aging on the ability to discriminate speech in

quiet, the older population generally experiences more

difficulty when speech is degraded (Schow et al., 1978) as

demonstrated by the common complaint of an older adult: "I

hear fine in quiet situations but I have difficulty when

there are several people talking at once or when it is


The decreased performance of the elderly on speech

tasks that have been altered or degraded has been well

documented (Bergman, 1971; Bergman, Blumenfeld, Cascardo,

Dash, Levitt, & Margulies, 1976; Calearo & Lazzaroni, 1957;

Kirikae et al., 1964; Konkle, Beasley & Bess, 1977; Korabic,

Freeman, & Church, 1978; Marston & Goetzinger, 1972; Smith &

Prather, 1971; Sticht & Gray, 1969). For example, Bergman

et al. (1976) conducted a 10 year longitudinal study of 282

adults ranging in age from 20 to 80 years. The subjects

were presented a battery of tests designed to assess the

perception of degraded speech. The CID Everyday Sentences

(Davis & Silverman, 1970) were presented under several

conditions: 1) undistorted and in quiet, 2) interrupted, 3)

at a rate 2.5 times normal rate, 4) filtered, 5) with

competing speakers, 6) increased reverberation, and 7)

overlapping words in isolation (similar to the SSW).

Results from the initial 282 subjects indicated a decrease

in performance for all conditions as a function of aging,

with the interrupted condition showing the greatest

decrement. Longitudinal testing, three and seven years

after the initial testing, yielded results which were

consistent with the initial findings. Bergman (1971)

concluded that the effect of age on speech understanding

ability is similar to that of reduced temporal integration.

That is, the central auditory system loses its ability to

integrate the signal as effectively.

The above findings are supported by a study by

McCroskey and Kasten (1980). McCroskey and Kasten (1980)

investigated the time needed for auditory fusion in three

subject groups: elderly subjects, learning disabled

children, and young normal subjects. The results indicated

that the elderly group required significantly greater time

to respond than either the learning disabled group or the

normal group. Kasten (1981) noted that 70-year-olds

required the same amount of time to complete the auditory

fusion task as a 3-year-old. McCroskey (1979) demonstrated

that temporal fusion is a function of age; more time is

required for children under 10 and adults over the age of


The decreased auditory performance on degraded speech

materials seen in the older population yet not evidenced in

the younger groups supports what Jerger and Hayes (1977)

refer to as the "central aging effects." That is, in

addition to the peripheral impairment associated with aging,

the higher auditory pathways may be involved (Bergman, 1971;

Jerger, 1973; Konkle et al., 1977; Kopra, 1982; McCarthy,


Based on the literature, therefore, it seems prudent to

include in the speech battery a test designed to assess

central auditory function. While filtered, interrupted,

time-compressed, and time-expanded speech have been used

previously to investigate central effects, the most commonly

used clinical tests reported are the Dichotic Digits Test,

the Staggered Spondaic Word Test and the Synthetic Sentence

Identification Test (Arnst, 1986). Of the three, the

Synthetic Sentence Identification (SSI) (Speaks & Jerger,

1965) is most commonly reported with the older population.

The SSI involves the presentation of 10 third-order

sentences (i.e. Small boat with a picture has become) in the

presence of a competing message (story of Davy Crockett) at

various message-to-competition ratios (MCR). The competing

message can be presented ipsilaterally (ICM) or contra-

laterally (CCM). Jerger (1973) generated complete

performance-intensity functions for both phonetically

balanced words and the SSI-ICM from 18 presbycusis adults, 5

young adults matched for hearing-impairment, and 5 normal

hearing adults. Comparison of the two functions indicated

that there was no significant difference between the groups

for maximum score obtained for the PB words (PBMAX), yet a

difference was evident when the PB words were presented at 5

dB SL (re: SRT); the presbycusis group demonstrated more

difficulty than the younger groups. Results of the SSI with

a 0 ICM, however, show a more pronounced effect. The

presbycusis group did markedly poorer on the SSI-ICM than

the young hearing impaired; the young hearing impaired

performance was poorer than the young normal group, and the

young normal group's performance also declined in comparison

to the PBMAX score.


Jerger and Hayes (1977) analyzed the patterns of the PB

and SSI-ICM functions and delineated specific patterns as

representative of 1) cochlear or peripheral involvement, 2)

eighth nerve lesion, and 3) central involvement. Central

involvement was determined to exist if the SSI function

falls below the PB function and could not be accounted for

by the audiometric configuration. Hayes and Jerger (1979),

using the SSIMAX PBMAX comparison, found that of 154 older

subjects (age 60 and older), 40% of the subjects

demonstrated primarily peripheral findings, 18% of the

subjects demonstrated mixed or intermediate findings, and

the remaining 42% were classified as centrally impaired.

That is, 42% of the older subjects were centrally hearing

impaired in addition to their peripheral hearing impairment.

Shirinian and Arnst (1982) also determined PI-PB and

PI-SSI functions on a population of elderly (age 60 85

years) hearing-impaired individuals classified as having

presbycusic losses. Unlike Jerger and Hayes's (1977) 20%

difference score between PBMAX and SSIMAX, Shirinian and

Arnst used a 14% difference score as the criterion for

centrally involved classification. While the results

supported the central aging effect described by Jerger and

Hayes (1977), the magnitude of the central deficit was not

as pronounced as the one reported by Jerger and Hayes


The studies by Jerger and Hayes and by Shirinian and

Arnst (1982) employed a 0 dB MCR. Orchik and Burgess

(1977) did not find a significant age effect between their

subject groups (10- to 12-year-old, 20- to 29-year-old, 40-

to 49-year-old, and over 60) at a 0 dB MCR. They suggested

that an MCR of -10 or -20 would be a more appropriate MCR

for revealing a central aging effect.


From the review of the literature on aging and hearing,

it is apparent that the older population is at risk for a

decrease in hearing sensitivity especially for the higher

frequencies. This hearing loss may also be accompanied by a

concomitant central auditory deficit, which becomes evident

primarily when the speech signal is degraded in some way.

However, it is important to note that not all elderly

individuals demonstrate a loss of hearing sensitivity and/or

central auditory dysfunction. Although the figures for the

prevalence of hearing impairment in the elderly vary as a

function of age, criteria for hearing loss, and the

population investigated, it is generally accepted that the

incidence of hearing impairment is approximately 30% and

rises as age increases (Harford & Dodds, 1982). Dalzell and

Puccia (1985) evaluated 250 patients admitted to a hospital

for intensive rehabilitation and reported that for the 65 to

69-year-old group only 44% had normal hearing sensitivity;

by the age of 85 only 11% had normal hearing.


It is equally important to note that of the population

with hearing impairments, not all have concomitant central

dysfunction. A review of two studies by Hayes and Jerger

indicated that in one study 18% (37 of 197) of the subjects

were classified as centrally impaired (Hayes & Jerger,

1979a) while in the other, 42% (65 of 154) were centrally

involved (Hayes & Jerger, 1979b). Shirinian and Arnst

(1982) reported an incidence of 74% for centrally involved.

Recently, Kricos, Lesner, Sandridge, and Yanke (1987), while

investigating hearing aid benefit as a function of central

auditory function, tested 24 subjects chosen semirandomly

from the clinic files at the Gainesville Veterans

Administration Medical Center. Of the 24 tested, 14 met the

criteria for classification of centrally impaired, yielding

a percentage of 58%.

This study was designed to investigate the central

auditory aging affect in an older population. A previous

established method, the SSI, was employed to classify

subjects as primarily peripherally impaired (PI) or

centrally impaired (CI). The subjects were then tested

using a more objective, electrophysiologic procedure known

as P3. It was hypothesized that the latency of the P3

component, in quiet, would correlate with the results of the

SSI; the latencies for the PI would be shorter than the

latencies for the CI subjects. This finding would suggest

that the CI subjects' increased difficulty with the SSI may


be attributed to an increase need in processing time. In

addition, the use of the noise condition for the P3 testing

would serve to further degrade the signal and result in

increased latencies for both groups.


This chapter includes a presentation of the various

methods and procedures that were used to investigate the

relationship between the latency of P3 and the central

auditory aging effect in the older, hearing-impaired adult.

More specifically, subject qualifications, audiometric and

electroencephalographic testing parameters and procedures

are presented.


Veterans were selected on the basis of age (65- to 75-

year-olds), pure tone thresholds, and driving distance, from

the audiology files at the Veterans Administration Medical

Center (G-VAMC), Gainesville, Florida, and were contacted

for participation via letters and/or telephone calls. Of

the 104 veterans contacted, 44 males agreed to participant;

31 of those satisfied the following conditions for inclusion

in this study: (a) pure tone thresholds no poorer than 35

dB HL (re: ANSI, 1969) for frequencies of 500 and 1000 Hz

for both ears; (b) eight or more (out of a possible ten)

correct answers on the Short Portable Mental Status

Questionnaire (SPMSQ) (Pfeiffer, 1975) indicating "Intact

Intellectual Functioning," (c) free from serious heart

disease, kidney disease, pulmonary disease with no history

of a stroke or cerebral vascular accident as assessed by a

medical case history questionnaire, and (d) live


Audiometric Testing

Audiometric testing was conducted in accordance with

ASHA guidelines and standard audiologic procedures. Pure

tone air (octave frequencies between 250 Hz and 8000 Hz) and

bone conduction (octave frequencies between 250 Hz and 4000

Hz) thresholds, and spondee thresholds were obtained for

each ear. Acoustic immittance testing, using a Madsen Z-

0174 immittance bridge, was conducted when air and bone

conduction thresholds differed by more than 10 dB at 2 or

more frequencies. Commercially recorded phonetically

balanced (PB) word lists (Auditec NU-6, Form A, List 1-4)

presented at 60, 70, 80, 90 dB HL (with contralateral

masking presented at 20, 30, 40, and 50 dBHL, respectfully)

were used to obtain performance intensity functions for

phonetically balanced words (PI-PB). The subjects were

asked to repeat the word that they heard. The maximum

score for the PB function was defined as PBMAX. The

Synthetic Sentence Identification (SSI) (Auditec recording)

Test with an ipsilateral competing message at 0 dB message-

to-competing ratio (MCR) was administered at the same

intensities as the PB words. The subjects were given an 8"

X 11" paper with the ten sentences typed in large print with

corresponding numbers (1 10). The subjects were informed

to call out the number associated with the sentence that

they had heard. The maximum score was obtained for the SSI

and defined as SSIMAX.

Audiometric testing was completed in approximately 60

minutes using a Grason Stadler two channel audiometer (GSI

10) coupled with TDH 49 earphones mounted in MX-41 cushions.

Testing was performed in a double room sound suite in the

Department of Speech Pathology and Audiology at the

Gainesville Veterans Administration Medical Center. The

equipment was shown to be in calibration prior to and at the

conclusion of the testing.

Electrophysioloqical Testing

Recording Conditions

Electrophysiologic recordings were made with a Cadwell

8400 data acquisition system. Silver/silver chloride

electrodes were fixed with electrode paste on the scalp at

Fz, Cz, and Pz, (electrode sites of the 10-20 system) and

referred to linked electrodes on each earlobe with forehead

serving as ground. Electrodes were placed supraorbitally

for the left eye and suborbitally for the right eye to

monitor eye movements. Electrode impedances were less than

5000 ohms. The electroencephalography (EEG) was amplified,

filtered through a bandpass filter (0.5 and 30 Hz), and

recorded over a 750 ms epoch. The EOG channel was averaged

to control for any time-locked EOG activity. Waveforms were

averaged on line by the Cadwell 8400 data acquisition system

and stored on floppy disks for future analysis.


The Cadwell 8400 data acquisition system controlled the

presentation of the stimuli following the classic "oddball"

paradigm. In this paradigm, a sequence of two tones are

presented with one tone occurring more frequently than the

other tone. The more frequently occurring tone is

designated as the "frequent" tone while the less frequent

tone is designated as the "rare" tone. The rare stimulus

occurs randomly at a probability of less than .50. For

example, if the probability for the rare stimulus is

determined to be .10 and a total of 300 tones are averaged,

270 of the tones would be the frequent stimulus and 30 would

be the rare stimulus. In other words, the frequent stimulus

occurs 90% of the time with a .10 probability. Separate

averages to the rare and the frequent signals are made and

are referred to as the common tracing (average of the

frequent stimuli) and the rare tracing (average of the rare


In this study, a 500 Hz tone was designated as the

frequent tone while a 1000 Hz tone was the rare tone. The

probability of the rare tone was .10 with a total of 300

tones presented. The tones had a 9.9 ms rise/fall and a 50

ms plateau time and were presented in a random series once

every .91 seconds. The tones were presented binaurally via

TDH 39 earphones with circumaural cushions at 50 dB nHL (66

dB SPL). Papanicolaou et al. (1985), Goodin, Squires,

Henderson, and Starr (1978), and Squires, Goodin, and Starr

(1979) have shown that latency of P3 is unaffected by

intensity until the intensity of the tones fall below a 15

dB Sensation Level.

Fifteen trials of the rare tone were collected and a

duplicate set of trials run again for replication. The P3

component was invoked in a quiet condition as well as in a

noise condition. In the noise condition, a 60 dB nHL (76 dB

SPL) medium band noise with energy between 200 and 2000 Hz

was presented binaurally through the earphones along with

the 50 dB nHL tones. The noise was determined to be at a

-10 dB signal-to-noise (S/N) ratio.

Testing was completed within one hour and was performed

in a room in the EEG Laboratory in the Department of

Neurology. The room was acoustically modified to reduce the

ambient noise floor. Carpet was placed under and behind the

Cadwell unit to reduce the noise emitted from the unit's

cooling fans. In addition, a free standing room divider was

placed as a dividing wall between the unit and the subject.

The ambient noise floor was assessed to be below 40 dBA at

the time of testing. The evoked potential system was

calibrated prior to and upon completion of testing.

Peak Identification

Although electroencephalographic activity was recorded

from three electrode sites, the Cz electrode site was used

as the primary site for determination of the peak latency.

When the component was difficult to define, the recordings

made at Fz and Pz were used for clarification purposes. Due

to the variety of waveform morphology, the latency of P3 was

determined by drawing lines paralleling the positive going

slope following the negative component referred to as N2 and

the negative going slope of P3. The latency was then

determined to be the point where the two lines of the slope

intersected. The latency was cursored to the nearest one-

hundredth ms.


Prior to audiological and electrophysiological testing,

the SPMSQ was administered and a medical case history was

taken. The SPMSQ was administered to ensure cognitive

intactness. The medical history was obtained to document

any history of cardiovascular, kidney, pulmonary disorders,

illnesses requiring hospitalization, otologic history and

medication records. Audiological testing consisted of pure

tone and spondee thresholds, PI-PB functions and PI-SSI

functions and were performed in that order. Testing was

terminated at this point for any subject failing the SPMSQ

and/or not meeting the given criteria for hearing


sensitivity (thresholds 35 dB HL or better for 500 and 1000


Subjects meeting the stated criterion proceeded to the

Neurology department for ERP testing. Subjects were seated

in a desk chair while electrodes were applied and then they

were moved to a recliner for testing. Following the

application of the electrodes, the subjects were given

instructions for the task. The subjects were instructed to

count mentally the number of high pitch tones and to

indicate when 5 or 10 rare (high pitch) tones had been

presented. The accuracy of their count was recorded by the

investigator for each condition. If the count differed by 3

or more, the subjects were reinstructed and that condition

was retested. The subjects were also instructed to relax,

avoid facial movements, keep their eyes closed, encouraged

not to blink and to mentally attend. The importance of

their maintaining their attention to the task was stressed.

Each subject was given a practice session to insure

understanding of the task. The practice session involved

the presentation of the stimuli until 1 or 2 rare tones were

heard. If the subject demonstrated difficulty with the

task, instructions were readministered followed by another

practice session.

Following the practice session, testing was initiated.

The quiet condition (QC) was presented first followed by the

noise condition (NC). A short rest period of approximately


1 minute was given between replications. Between the QC and

the NC, the earphones were removed and the instructor

engaged the subject in a brief conversation to maintain

mental alertness.

Upon completion of the testing, the difference between

the subject's PBMAX and SSIMAX was computed (defined as the

difference score). If the difference score was 16 or

greater in either ear the subject was classified as

centrally-impaired (CI). If the difference score was less

than 16 for both ears, the subject was considered to have

primarily a peripheral impairment. Subject selection

continued until at least 15 subjects existed in each group.

This chapter includes a presentation of the

experimental findings investigating the relationship between

the P3 component and central auditory function in an older,

hearing-impaired adult population. Subjects were classified

as having a central or a peripheral impairment through the

use of performance intensity functions for phonetically

balanced words (NU-6) and the Synthetic Sentence Identi-

fication (SSI) test. The latencies of the P3 component

elicited in a quiet and a noise condition were correlated

with these groups. A general discussion of the findings and

their implications will follow in Chapter V.

Thirty-one subjects met the criteria for inclusion in

this study. Fifteen subjects were considered to have

primarily a peripheral impairment (PI) based on a difference

of less than 16 for both ears between the PBMAX (maximum

score on the phonetically balanced word lists) and SSIMAX

(maximum score on the Synthetic Sentence Identification

Test). Sixteen subjects were classified as "centrally

impaired" (CI) based on the criterion of a DIFSCORE (the

difference of the SSIMAX from the PBMAX) of 16 or more for

either ear. Of the 16 subjects in the CI group, 5 subjects

demonstrated a mixed impairment. That is, four of the

subjects demonstrated central impairment for the left ear

and a peripheral impairment for the right ear. The fifth

mixed impairment subject had the reversed situation. These

five subjects form a third classification as the mixed-

impairment (MI). Discussion of their results will be

included in the CI group except when noted.

The ages of the two groups were tightly controlled, with

both groups ranging in age from 65- to 75-years-old. The

mean age for the CI group was 68.9 years (SD = 3.0) and was

not found to be significantly different from the mean age of

68.2 years (SD = 2.8) for the PI group through an one-way

analysis of variance (ANOVA) (F (1, 29) = 0.49, p = 0.4896).

Behavioral Measurements

Differences Between Ears

The means and standard deviations for the behavioral

measurements, pure tone average (PTA), spondee threshold

(ST), maximum percentage for the PB words (PBMAX), maximum

percentage for the SSI (SSIMAX), and the difference between

the PBMAX and SSIMAX (DIFSCORE), for the two groups,

classified as CI or PI, are presented in Table 1. Five two-

way ANOVAs for repeated measures indicated that the right

(RE) and left ear (LE) were not statistically different (p >

.05) within groups for any measurement.

Table 1
Means and Standard Deviations
for Behavioral Measurements


RE Mean 29.4 21.7 86.0 53.7 34.1
SD 8.1 6.9 14.5 30.7 23.9
LE Mean 29.7 21.4 84.2 47.5 36.7
SD 8.8 6.3 14.9 29.1 19.2

RE Mean 22.3 17.9 93.9 91.3 3.7
SSD 9.2 8.0 13.1 15.1 3.7
LE Mean 21.1 17.6 94.8 94.7 2.0
SD 10.1 7.8 8.0 9.1 3.4

a n = 16 bn = 15


It must be noted, however, that the means and standard

deviations in Table 1 are according to group classification.

Recall that five of the CI subjects had one peripherally

defined ear and one centrally defined ear so that in the

case of the CI group right ear, the results are computed on

data from four ears technically qualifying as peripheral

ears. For the left ear, the data include only one

peripherally defined ear. To clarify the results, the data

were separated into three groups, PI, CI (n = 11), and MI.

The means and standard deviations for the behavioral

measurements for the three groups are shown in Table 2.

When the subgrouping for the CI group is used, the CI will

be designated as CI-11, indicating that the grouping is

different from the original CI group (n = 16) and consists

of eleven subjects. To test whether this subgrouping would

show an ear effect, the right ear data were compared to the

left ear data for each group. Because of the unequal sample

sizes, a series of t-tests for dependent variables were

performed (using the Epistat computer program for PC) (Table

3). No significant differences between ears were noted for

any groups (p > .05).

Differences Between Groups

An ANOVR for between subjects effects was used to

determine significance between groups CI and PI for the

behavioral measurements. Figure 3 demonstrates the obtained

means for the behavioral measurements between groups for the

Table 2
Means and Standard Deviations
for Behavioral Measurements
of the Three Groups


RE Mean 28.6 19.9 85.1 40.0 46.0
SD 9.3 7.4 14.4 26.8 18.3
LE Mean 29.6 22.1 81.5 38.2 43.3
SD 8.9 5.8 15.2 28.9 18.7

RE Mean 31.2 25.6 88.0 84.0 4.0
SSD 4.9 3.8 16.0 8.9 10.7
LE Mean 29.8 20.0 90.4 68.0 22.4
SD 9.7 7.9 13.4 17.9 11.3

RE Mean 22.3 17.9 93.9 91.3 3.7
SD 9.2 8.0 13.1 15.1 3.7
LE Mean 21.1 17.6 94.8 94.7 2.0
SD 10.1 7.8 8.0 9.1 3.4

a n = 11
Sn = 15
c n = 15

Table 3
Results of t-tests
Between the RE and LE ear
Within Groups

Measurement t statistic p value

PTA .4848 .64
ST 1.1418 .28
PBMAX 1.0471 .32
SSIMAX .4822 .64
DIFSCORE .5435 .60

PTA .3215 .76
ST 2.4188 .07
PBMAX .2308 .83
SSIMAX 1.5540 .19
DIFSCORE 2.0201 .11

CI-1l n = 11;

MI n = 5

94.8 94.7















Figure 3.


The means for the pure tone averages (PTA),
spondee threshold (ST), maximum score for PB
words (PBMAX), the maximum score for the SSI
(SSIMAX), and the difference between the PBMAX
and SSIMAX (DIFSCORE) are presented for the left
ear for the PI and the CI groups. The PTA and
ST are expressed in decibels while the PBMAX,
SSIMAX, and DIFSCORE, are percentages.




left ear. For the PTA measure, the PI group had better

thresholds than the CI group for the RE (22 dB HL for the PI

and 29 dB HL for the CI) and the LE (21 dB HL for the PI and

29 dB HL for the CI). The observed significance level for

the RE was 0.03 and 0.02 for the LE. For the ST scores,

neither the RE (E = 0.16) nor the LE (p = 0.14) showed a

difference between the CI and PI group; the ST was 17 dB HL

for the RE and LE for the PI group and 21 dB HL for both

ears for the CI. The PBMAX scores for the CI group were

lower than the PI group's and showed a significance between

the groups for the LE (E = 0.02), yet failed to show

significance for the RE (p = 0.12). The SSIMAX scores were

significantly different between groups; RE observed

significance level was (p = 0.0002) and LE observed

significance level was (p = 0.0006). Likewise, the

DIFSCORE, as expected because of subject grouping criteria,

showed significant differences beyond the 0.0001 level of


Subjects were grouped according to the PB-SSI

difference score; therefore, significant differences were

expected between the groups for the SSIMAX and the

difference score. Likewise, note the differences (Table 1)

in the standard deviations (SD) between the groups for the

SSIMAX, and for the difference scores. While the SSIMAX

scores do not violate the assumption of normal variance, the

DIFSCORE does. Categorization into the PI group was based


on a difference score of less than 16, thus the range was 0

to 16 (actual range was 0 to 12). Categorization into the

CI group was based upon a difference score of 16 or more,

yielding a range of 16 to 100 (actual range was 10 to 88).

The range of the PI group was more limited in respect to the

CI group, accounting for the differences in the standard

deviations of such measure.

There was also a difference between groups for the

three frequency pure tone average (3-PTA). Further

examination of the data indicated that there were no

statistical differences for thresholds at 500 and 1000 Hz

yet there was a significant difference for 2000 Hz. The

poorer thresholds for 2000 Hz for the CI group thus affected

the 3-PTA. It must be noted, however, that thresholds at

500 and 1000 Hz were controlled, with the investigator

eliminating any potential subject with poorer than 35 dB HL

threshold at either frequency. Thresholds at 4000 Hz were

also statistically poorer for the CI group than for the PI

group (Table 4).

Again, it must be noted that the data presented in

Table 4 are based on group classification of CI and PI. To

determine if the data from the MI group were influencing the

group differences, the CI group was again subdivided into MI

and CI-11 groups and submitted for statistical analysis.

A series of t-tests for independent variables indicated

that there were no differences between the MI and CI-11 for

Table 4
Means and Standard Deviations
for Pure Tone Thresholds
Between Groups

Group 500 1000 2000 4000

CI Mean 19.4 23.7 47.2 77.5
SD 7.5 8.8 15.8 20.1

PI Mean 15.0 17.7 28.0 51.7
SD 7.8 9.2 12.9 20.6

t-test 1.59 1.87 3.68* 3.54*

* significant at .01 level

PTA, ST or PBMAX. There were, however, significant

differences between the groups for the SSIMAX and the

DIFSCORE; the SSIMAX was poorer and the DIFSCORE was greater

for the CI-11 group (Table 5).

Summary of Behavioral Results

Results of ANOVRs and t-tests suggest the following:

1. There was no difference in age between the groups

(CI and PI).

2. The two groups were similar in terms of hearing

sensitivity for 500 and 1000 Hz and for spondee thresholds.

The CI group exhibited poorer high frequency hearing


3. The two groups did not differ in their ability to

understand monosyllabic words (PBMAX).

4. The two groups differed in their performance on the

SSI, resulting in statistically different SSIMAX and

DIFSCORE. This difference is a factor of the original

grouping criterion.

5. The MI group did not differ from the CI-11 group for

any measurement, except the SSIMAX and DIFSCORE.

Table 5
Results of the t-test
for Behavioral Measurement
Between the CI group and the MI group

Measurement t statistic p value

PTA .3215 .7639
ST 1.5983 .1323
PBMAX .3620 .7227
SSIMAX 3.5199 .0034*
DIFSCORE 4.4083 .00005*

PTA .0332 .9740
ST .5977 .5596
PBMAX 1.1262 .2790
SSIMAX 2.1064 .0537**
DIFSCORE 2.2856 .0384**

* computed with data from the 4 subjects that were
classified as PI for the RE.

** computed with data from a MI subject with a LE PI.

df = 14

Electrophysiological Results

Identification of the P3 Component

A representative event-related potential response, with

replication, elicited in quiet, from one subject is shown in

Figure 4. The top three tracings are the "common tracing"

(e.g. average response of the frequent tones) with the top

tracing recorded from the scalp location of Fz, the middle

tracings from Cz and the bottom tracing recorded from Pz.

The bottom three tracings are the "rare tracings" (average

response of the rare tones). The same electrode montage was

used for the rare tracings with the top tracings recorded

from Fz, the middle tracing from Cz, and the bottom tracing

from Pz. Note the large positive component occurring after

the second positive peak in the rare tracings only. This is

the component labeled P3. While the morphology of this

component varied between subjects, it was easily identified

by using the robust components of P2 and N2 as additional

markers. Several examples of the wave morphology of the P3

component are presented in Figure 5. Note the well-defined,

sharp positive component in Tracing A compared to Tracing B,

which is characterized by two positive peaks. These peaks

have been referred to in the literature as P3a and P3b and

are considered to be subcomponents of P3. For subjects

demonstrating both a P3a and P3b component, the latency of

P3 was determined by using the up-going slope of P3a and the

down-going slope of P3b. Regardless of the morphology, the


Figure 4.

- F z
_.--^ '^- ^ ^.C z
7c^ ~-JP z
4 P3

\/' VC z


I5 uv

375 750 ms

An example of the P3 component
waveform from one subject. The
top three tracings are the average
of the frequent tones. The bottom
three tracings are the average of
15 rare tones for each replication.
Fz is the recording from the fron-
tal electrode, the Cz is from the
vertex electrode and the Pz is
from the parietal electrode.


Figure 5.

/^ Cz

f5 uv N Pz

375 750 ms

Representative response wave-
forms of the rare tracings only
from two subjects demonstrate
the variability in the P3
component. Note that in Tracing
identified and characterized by
a well-defined, sharp positive
peak whereas in Tracing B, the
P3 component is characterized by
multiple peaks. These peaks are
labeled as P3a and P3b.
A th P31 u ce i V

peak whereas in Tracing B, the
P3 component is characterized by
multiple peaks. These peaks are
labeled as P3a and P3b.

P3 component was easily identified for all subjects in the

quiet condition.

The identification of P3 was somewhat more difficult in

the noise condition. The entire waveform morphology was

altered by the introduction of the noise. The morphology

tended to become "noisier" with the P3 component flattening

and widening (Figure 6). This widening of P3 is most likely

the result of averaging "quick" responses with "delayed"

responses. That is, in some cases, the subject detected the

rare tone rapidly, hence a shorter latency P3 and at other

times, the detection was more prolonged resulting in a

delayed P3. The averaging of the early and late individual

P3s resulted in a broader component. The P2 component also

became somewhat less distinct. In cases where P2 was not

obvious, N2 (the second negative component) was used as the

landmark for the onset of P3. As in the case of the quiet

condition, P3 was identifiable in all subjects.

Latency of the P3 Component

As previously stated the latency of P3 was determined

by drawing lines paralleling the slopes of the component and

placing the cursor at the point the lines intersected. The

waveforms were recalled from the floppy disk and displayed

on the monitor. The lines of the slopes were drawn with

grease pencil on the monitor and then cursored.

The latencies for the quiet and noise conditions for

the groups were submitted to the General Linear Models







375 750 0 375

Figure 6.

The tracings on the left are responses from the quiet condition for one
subject. Note the well defined peaks. The tracings on the right are the
response invoked in noise from the same subject. Note the deterioration of
the well-formed peaks and the widening of the P3 component. The vertical
line indicates the latency of the P3 component. The tracings at the extreme
bottom of the graphs are the tracings recorded from the eye monitor


Procedure from the Statistical Analysis System (SAS)

program. Results of the univariate tests of hypotheses for

within subject effects indicated a significant interaction

between groups and conditions (Table 6). Because of this

interaction, the significant main effects of condition were

not considered. The P3 component was shorter (330.78 ms)

for the PI group than for the CI (354.16 ms) group (F (1,

29) = 6.60, E < 0.02) for the quiet condition. While the

null hypothesis of no difference between the quiet and noise

conditions could be rejected at the p < 0.0001 significance

level, there was no statistical difference between the means

of the groups in the noise condition. The mean latency of

the CI for noise was 383.20 ms and was not significantly

different from the mean latency of 393.54 ms for the PI

group (F (1, 29) = 0.98, p = 0.3304). This may be more

easily demonstrated by Figure 7.

The relationship between the noise and quiet conditions

for each group was further investigated using the Pearson's

product-moment correlation coefficient. There was a highly

significant relationship for the PI group (r = .76, p =

0.00009). For the CI group, there appeared to be a pattern

as indicated by eleven of the group showing little or no

change in the latency from the quiet to the noise condition,

The other five (not the MI group), however, had shifts in

latencies from 50 to 100 ms resulting in correlation of r =

.12 (p > 0.05).

Table 6
Summary of the ANOVR for the
P3 Latency Measures

Source Degree of Sum of Mean F Value p Value
Freedom Squares Square

Group 1 658.5 658.5 0.63 0.4349

Condition 1 32619.7 32619.7 75.03 0.0001

Group*Condition 1 4400.7 4400.7 10.12 0.0035

ERROR 29 12607.2 434.7


400 ms











- .

- &i- B



Figure 7.

The mean latencies for the
quiet and noise conditions
for the PI group and the CI
group. The means are signi-
ficantly different for the
quiet condition between
groups whereas the means for
the noise condition are not
different between groups.
Between the quiet and noise
conditions, the latencies
are different for both the
PI group and the CI group.



: t


* ,4, ,
: :: : i

* , ,

: :

. : : : :


The P3 Component for the Mixed Losses

The five subjects in the MI group were recalled for

monaural testing to determine if there was an ear effect for

the P3 component. They were retested for the quiet

condition following the original procedures with the

exception that the stimulus was presented to one ear at a

time and then to both ears. The order of presentation was

binaural, right ear and finally, the left ear. Testing was

conducted for the quiet condition for 4 of the subjects.

The other subject was retested for both the quiet and noise

conditions. A representative waveform from subject #135 is

shown in Figure 8. Only the responses evoked by the rare

tones are shown. Note the similarities in the waveform

morphology between the top three tracings recorded from the

right ear (AD), left ear (AS), and binaural (AU). Results

of ANOVR indicated no statistical differences for the

latency measures between ears and between monaural versus

binaural (F (2, 9) = .32, p = .73).

Stability of the P3 Component

To analyze replicability and stability of the P3

component, five subjects were recalled for repeat testing.

Four subjects were retested for both quiet and noise

conditions. The fifth subject received only the quiet

condition due to time restrictions. The time between

testing ranged from 24 days to 105 days (mean of 86 days).

Waveforms from the initial testing as well as the repeat

Cz AD,", ,'' ,I A". Dj
I .I / ',,I ,,. ,',. fl' .... , .

Iz 1 /, '' 1, -
z .. . .' '

Pz r ,J .. ,

Fz -' -'
C z In ,--."."_, r ,,, I)

"j- 5 uv

Figure 8. A representative waveform demon-
strating the recorded responses from
stimulus to the right ear (AD) only,
left ear (AS) only, and to both ears
(AS). Replications of these
responses were not done. Note the
similarity between the tracings.


testing are shown in Figure 9. The waveforms and the

latency of P3 remained quite stable over the period of

several months. A Pearson product-moment correlation

coefficient, r = 0.97 (p = .0002), indicated a strong

correlation between mean latencies of the initial test

session and the repeat session.

Subject 107


7 Dec. 1987

17 Mar. 1988

Subject 115

Subject 102



11 Dec. 1987

24 Mar. 1988


20 Nov. 1987

14 Dec. 1987

750 ms

Figure 9. Waveforms from three subjects recorded on
two separate occasions show the stability
of the P3 component.


In this chapter, a discussion of the relationship that

was found between the central auditory status and the P3

component in older hearing-impaired adults is presented.

The purpose of this investigation was to determine if the

event-related potential known as P3 could be used as a

measure of central auditory function in the older adult.

Thirty-one male veterans were administered a battery of

behavioral tests and the P3 event-related potential (ERP)

test, the latter under two conditions, in quiet and noise.

The subjects were classified as peripherally impaired (PI)

or centrally impaired (CI) based on the difference between

the maximum score for the Synthetic Sentence Identification

test (SSI) and the maximum score for discrimination of

phonetically balanced word (NU-6) lists. Sixteen men were

identified as having a central component while the other

fifteen were defined as having primarily a PI, and the

grouped data were submitted for statistical analysis. The

results will be reviewed and discussed with reference to

prior research. Implications and limitations of this study

as well as the need for future study will be addressed.

The P3 Component in Quiet

The latency of the P3 component for the PI group, in

quiet, was found to be consistent with previously reported

age-related norms for auditory stimuli of easy

discrimination (Brown, Marsh, & LaRue, 1983; Goodin,

Squires, Henderson, & Starr, 1978; Pfefferbaum, Wenegrat,

Ford, Roth, & Kopell, 1984; Picton, Stuss, Champagne, &

Nelson, 1984; Polich, Howard, & Starr, 1985b; Squires,

Chippendale, Wrege, Goodin, & Starr, 1980). For example,

Polich et al. (1985b) reported that the mean latency for the

P3 component was 329 ms for 60- to 69-year-olds. Squires et

al. (1980) computed a rate per year increase in latency as a

function of age and task difficulty and determined that the

latency increased .79 ms per year for easy tasks. Using the

Squires et al. (1980) rate and the base mean of 287 ms (the

mean for 20-year-olds suggested by Goodin et al., 1978), the

latency for the age group of 65- to 75-year-olds was

computed to be 327 ms, just three milliseconds shorter than

the obtained latency of 330 ms for the PI group. The

latency of the P3 component for the PI group thus shows a

normal progression as a function of age.

On the other hand, when comparing the Goodin et al.

(1978) latency of 329 ms or the computed latency of 327 ms

to the CI group, the results are not consistent with a

normal age effect. The obtained latency for the CI group

was 354 ms and was significantly longer than the 330 ms


latency of the PI. This suggests that the CI group required

longer time for stimulus evaluation. The basic premise of

the P3 component is that it is a measurement or reflection

of the timing of stimulus evaluation and can be used as a

relative indicator of how long it takes to evaluate a

stimulus. Hence, longer latencies indicate longer

processing time. The results from this study thus seem to

support this premise.

To further support the implications of the CI findings,

the expected latency can be computed for the CI as was

computed for the PI group. Recall, that the computed

latency using the rate per year for the easy task was

determined to be 327 ms for this age group; substantially

shorter than the CI latency of 354 ms. However, if the

latency is determined by using the rate per year that

Squires et al. (1980) reported for the more difficult task

(1.49 ms per year), the expected latency would be 361 ms; a

value much more in line with the obtained latency for the CI


The electrophysiologic results obtained in quiet thus

support the behavioral results. That is, the subjects who

had little difficulty identifying the test message from the

competing story on the SSI, i.e., the PI group, also had

little difficulty evaluating the stimuli for the P3 testing

in quiet. The subjects who performed more poorly on the

SSI, i.e., the CI group, demonstrating more difficulty

separating the message from the competing story, also

required additional stimulus evaluation time of the P3


Thus, the P3 component, evoked in quiet, reflects an

increased processing time for the older adult. The latency

of the P3 component further demonstrates an even greater

need for stimulus evaluation time for those subjects defined

as centrally impaired. The results seem to support the use

of the P3 component as a measure of central auditory

function and indicate that the P3 measurement and the SSI

test assess similar functions in the older adult.

The P3 Component in Noise

The latency of the P3 component, repeatedly, has been

shown to be a function of the complexity of the stimulus;

the more complex the stimulus, the longer the latency.

Ford, Duncan-Johnson, Pfefferbaum, & Kopell (1982), however,

reported that it was not the difficulty of the decision

(i.e. increasing the memory demand) that influenced the P3

latency but the difficulty of discrimination of the task.

The results of this study support that finding. The latency

of the P3 component evoked in the presence of noise resulted

in significantly longer latencies for both the CI and PI

groups when compared to the quiet condition. By introducing

a background noise, the discrimination between the rare and

common tones became more difficult which in turn increased

the stimulus evaluation time and hence, the latency of P3.

The introduction of the noise, however, affected each

group differently. Originally, it was proposed that the

introduction of the noise would increase the complexity of

the task for both groups resulting in approximately equal

shifts in latency. The introduction of the noise resulted

in increased latencies for both the groups, yet the

increases were not equivalent. The PI group latencies

shifted from 330 ms in quiet to 393 ms in noise, a 63 ms

shift as a function of noise. The CI group latency shift

between conditions was only 29 ms; 354 ms to 383 ms. While

the shifts in latencies, between quiet and noise, were shown

to be significantly different for each group, the latencies

between groups for noise were not shown to be significantly

different. In effect, the noise served to increase the

complexity within bounds. That is, the introduction of the

medium band noise with energy around the test stimuli,

resulted in increasing the stimulus evaluation time, as

indexed by the P3, to approximately 390 ms for 65- to 75-

year-olds regardless of the impairement (PI or CI). The

unequal shifts in the P3 component then, suggest that the

noise only slightly interfered with the processing of the P3

stimulus for the CI group and greatly interfered with the PI

group's stimulus evaluation time. Perhaps, since the PI

group had a more intact central system than the CI group,

the central system became more impaired with the

introduction of the noise, whereas the CI group, with their

already existing central impairment, was less "impaired" by

the noise.

Interestingly, when waveforms for the conditions were

superimposed, that is, the quiet waveform superimposed on

the noise waveform, for each individual, a trend appeared.

For the PI group, there was a definite shift in the P3

component from the quiet to the noise condition with the

shift ranging in time from 45 ms to 131 ms. Note the

obvious shift of the P3 component for a PI subject in Figure

11 (Tracing A). In fact, a shift (ranging from 45 ms to 130

ms) occurred for all fifteen PI subjects. Now compare the

PI results to the superimposed waveforms (Tracing B) from a

CI subject. Note that there was no obvious shift. Eleven

of the sixteen subjects of the CI group (these included

subjects from both the MI group and the CI-11 group) showed

no shift in the P3 latency as a function of conditions. The

five CI subjects showing shifts in latencies, had shifts

ranging from 43 ms to 90 ms. A closer examination of the

behavioral data failed to show any consistency between

measures to account for whether or not the subject's P3

component shifted as a function of the noise. Table 8

reports the behavioral and P3 results for the left ear of

the CI subjects. Note the wide variability for the P3

component and the behavioral measurements.

0 375 750 ms

Figure 10. Tracing A represents the single
recorded reponses from a PI subject,
demonstrating an obvious shift in
the P3 component. Note the earlier
occurrence of the P3 component in
the quiet condition (. .. .) when
compared to the noise condition
(-----) for the PI subject.
Tracing B represents the reponses
from a CI subject. Note that the
P3 component did not shift as a
function of the conditions.

Table 8
Behavioral and P3 Data
for the CI Group


103 70 28 25 88 60 28 368.75 365.62 N
107* 72 32 25 100 90 10 343.75 359.37 N
111 74 34 25 92 70 22 384.37 396.87 N
115* 66 18 15 96 80 16 359.37 365.62 N
119* 66 23 10 100 70 30 407.25 418.75 N
122 67 37 35 76 60 16 328.12 328.12 N
127 69 42 25 64 0 64 378.12 381.25 N
129 67 33 25 64 0 64 359.37 356.25 N
135* 66 33 20 88 50 38 365.62 393.75 N
141 65 20 18 96 50 46 375.00 375.00 N
144 69 17 18 84 10 74 353.12 368.75 N

102* 70 43 30 68 50 18 318.75 362.50 Y
104 75 20 20 96 60 36 328.12 415.62 Y
113 67 25 15 92 50 42 353.12 415.62 Y
137 72 43 22 52 0 52 328.12 415.62 Y
143 68 27 15 92 60 32 315.62 412.50 Y

* subjects within the MI group

Results of the noise condition, therefore, indicated

that the introduction of the noise created a more difficult

task for both groups, with the PI group showing a much more

systematic effect than the CI group. Specifically, a

bandpass noise centered around the stimulus frequencies can

function to mask the test frequencies without being too

obnoxious for the listener. In addition, this type of noise

will not be greatly affected by the hearing sensitivity, or

lack of, at other frequencies; a concern when dealing with

the older population. A -10 signal-to-noise ratio, as used

in this study, is an appropriate level of noise presentation

for the purpose of demonstrating more of an effect. Polich

et al. (1985a) introduced a white noise at a 0 signal-to-

noise ratio and recorded the difference for the P3 component

between quiet and noise. The reported shifts in latency

were shown to be of a magnitude of 8, 10, and 11 ms.

Although they reported significance for these shifts, it is

questionable how clinically significant an 8 ms shift is

based on the variability of the latency data. The more

intense noise yields larger, more observable shifts in

latency, lending more confidence that a shift indeed


The Mixed Group

An interesting finding in this study that was not

originally considered was the presence of a third group of

subjects. Five of the subjects in this study formed a third

group by demonstrating one PI ear and one CI ear. Although

the percentage (31%) of the mixed impairment (MI) subjects

in this study was somewhat higher than has been previously

reported (Shirinian & Arnst, 1982), the existence of mixed

impairments has been documented. Shirinian and Arnst

(1982), using the same method of classification of CI or PI

as was used in this study, investigated the central effect

as a function of ears. They reported that 17% of their

subjects exhibited one central ear and one peripheral ear.

While the presence of a mixed impairment was fairly

consistent with previous literature, the ear exhibiting the

central component was not. For all but one of the mixed

group in this study, the left ear showed the central effect

while the right ear was defined as PI. Shirinian and Arnst

(1982) noted that their subjects showed a right ear central

effect and the left ear was more consistent with peripheral

findings. Likewise, Jerger and Hayes (1977) compared the

means of the PBMAX and SSIMAX between ears (n = 67; age 60

to 69) and reported that the right ear showed more of a

central effect than the left ear. Jerger and Hayes (1977)

noted that their findings were consistent with previous

research using filtered words (Palva and Jokinen, 1970) and

dichotic consonant vowels (Speaks, Niccum, and Carney,

1977); both showing right ear effects.

Likewise, the right ear predominance was not supported

in this study for the eleven "pure" CI subjects. In fact,

the CI-11 group failed to show any ear predominance.

Although the absolute DIFSCORE for seven of the eleven

subjects was greater for the right ear, a t-test for

dependent variables showed no significant difference ( =

.37, E = .71) between the means of the right (45.0) and left

(43.2) DIFSCORE. These findings may be accounted for by the

small sample size for both the MI group (n = 5) and the CI-

11 group (n = 11).

Returning to the issue of the MI group and the fact

that behaviorally, one ear was defined as CI and the other

was defined PI, the results of the electrophysiologic

testing are examined. One might have expected that if an

ear effect existed for the P3 component, results from each

ear would be different; the results from the central ear

would be consistent with the CI group's results and vice

versa. Recall that the ANOVR indicated no significant

latency differences between the presentations: right ear

only, left ear only, and binaural. Examination of the

waveforms also support the similarity between the ears. The

finding that each ear behaves similarly suggests that either

the peripheral component or the central component overrides

the other impairment. Examination of the MI data reveals a

latency value for P3, in quiet, consistent with the latency

value of the CI group. It would appear, then, that for the

MI group the central impairment overrides the peripheral

impairment for the P3 component. The consistency of the

latency values between the MI group and the CI group

support the central origin of the P3 component.

In summary, there seems to exist a percentage of the

older hearing-impaired population that have a mixed

impairment as evidenced by the behavioral results. The

results from the P3 testing suggest that the P3 component is

influenced by the central nervous system intactness and is

influenced to a much smaller degree by the peripheral

impairment. The central impairment appears to override the

peripheral impairment, however, to what extent has yet to be

determined and warrants further investigation.

The P3 Component: Summary

The results of this investigation indicated two

significant findings: (1) The latency of the P3 component

invoked, in quiet, can be used as a measure of central

auditory function in the older, hearing-impaired population

and (2) bandpass noise is a useful means of increasing the

complexity of the discrimination task. The findings failed

to support the use of the noise condition alone as a measure

to delineate the two groups, the PIs and the CIs.

The P3 component is an index of stimulus evaluation

time. As Duncan-Johnson and Donchin (1977, p. 467) stated

the "... P3 ... manifests on the scalp a process intimately


involved in the processing of information...." The results

of this study support that. For the quiet condition, the

latencies of the PI group were consistent with the

previously reported age-related norms. The CI group's

latencies were significantly longer than the PI, again,

indicating a longer processing time. The subjects in this

study were all healthy, cognitively intact older adults.

The results of the P3 would then suggest that the difficulty

that the CI group demonstrated understanding speech, as

assessed by the SSI, was due to slower processing time. It

appears that P3 and the SSI index approximately the same

processes. That is, the ability to succeed on the SSI

necessitates the ability to rapidly process the stimulus,

evaluate the relevant information, and ignore unwanted

messages. The P3 component is also affected by the ability

to rapidly process and evaluate the relevancy of the

information. Therefore, the P3 component is a viable

measure of central auditory processing.

Likewise, the noise has been shown to be an effective

and easy method to increase the difficulty of the

discrimination of the stimuli. Both groups were affected by

the noise. The noise, however, failed to delineate the two

groups as being different if the latency for the noise

condition was used alone. There was a large degree of

intrasubject variability so the use of absolute latency may

be tentative. More definitive however, is the use of the P3


component shift from quiet to noise. Recall, that eleven of

the CI group demonstrated no observable shift in the P3

component; all fifteen of the PI group had obvious shifts in

the P3 component. Therefore, if a client demonstrated no

observable shift in the P3 component, it would be with

confidence that a classification of central impairment could

be made. Unfortunately, the reverse is somewhat more


Wide intrasubject variability was noted for both the

quiet and noise conditions. Yet, this variability is

consistent with previously reported findings (Pfefferbaum et

al., 1984; Howard & Polich, 1985) and may be explained by

the possible heterogeneity of the older population. It is

well known that the older adult population is highly

heterogenous; everyone ages yet the rate and degree of aging

occur differently across the population. For example, as

the central nervous system ages a number of changes occur.

There is a loss of neurons and their processes (Brody, 1955,

1985; Kenny, 1988; Shock, 1983). Along with the loss of

neurons, the remaining cells may become impaired from

deposits of fat or lipofuscin. The surviving cells also

form new connections to compensate for the lost ones, thus

"rewiring" the circuitry. In addition, there is a loss of

myelin resulting in a reduction of the conduction velocity

time, or the speed at which action potentials travel. It is

important to note, however, that these changes do not occur

uniformly within the brain nor across people. It may be

possible that the CI subjects are demonstrating this non-

uniformity of age-related changes and the latency of the P3

component, invoked in the quiet and noise, is reflective of

this. It could be suggested that the subjects in the CI

group demonstrating shifts of the P3 component in noise, a

characteristic more in line with the PI findings, may have

less of a central component than the individuals with no

shift. Yet if this is the case, the DIFSCORE would be

expected to be small or relatively closer to the criterion

of a DIFSCORE of 16. The SSI results do not support this.

In fact, subject 137 had DIFSCOREs of 60 (RE) and 50 (LE)

and a latency shift of 87 ms. Other subjects demonstrating

no shift in the P3 components had DIFSCOREs ranging from 16

to 74. Perhaps, this finding indicates the dissimilarity of

the P3 component and the SSI test. That is, perhaps the P3

component is influenced by a factor that has little affect

on the SSI or vice versa. It may be possible that the P3

component may be more sensitive and more reflective of the

true status of the central auditory system. A definitive

answer to this speculation awaits further investigation.

One may question, however, how a procedure that uses

"tones" as the stimulus can adequately assess the central

auditory system. Bocca and Calearo (1963) stressed that

tones are not an appropriate stimulus for central auditory

testing because the tones do not reduce the redundancy

sufficiently. More recently, Bosatra and Russolo (1982)

supported Bocca and Calearo's precept. A battery of central

tonal tests (auditory lateralization, temporal order,

auditory pattern) and central speech tests (sensitized

speech and the SSI) were administered to 50 healthy subjects

age 60 to 80 years old. The results demonstrated that the

central speech tests showed the central auditory aging

effect while the central tonal tests results were equivalent

to the norms of younger adults. The use of tones for the P3

component, then may seem questionable. However, the P3

component is an endogenous potential and as an endogenous

component, it is not influenced by stimulus parameters such

as intensity, rise/fall time, or whether the stimulus is in

the auditory, visual, or somatosensory modality. Instead,

the P3 component is influenced by the motivation, attention,

and cognitive skills of the subject. Thus, the P3 component

can be invoked with similar results using speech or tones.

In addition, the use of tones has several advantages

over speech. When using speech, the peripheral auditory

system's influence on the discrimination of the stimuli must

be considered. With peripheral involvement, the likelihood

for poorer discrimination ability is increased. This

increased difficulty in discriminating speech may be a

function of greater peripheral involvement and not

necessarily a central factor. Yet the inability to

discriminate speech at the peripheral level may influence

the perception of speech at the central level, thus

possibly influencing the P3 component. The use of tones

eliminates that problem. As long as the presentation level

of the tones is 15 dB above threshold for that frequency,

the P3 can be invoked and the latency of the P3 component is

unaffected by the intensity (Goodin et al., 1978). Again,

since the P3 is an endogenous component, it is independent

of the test frequencies and results will be consistent if

low frequencies or high frequencies are used. Therefore,

the physical characteristics of the stimulus will have

little effect on the P3 testing when using tones.

In summary, the P3 component, invoked in quiet, is a

viable method for assessing the central auditory status of

the older adult, hearing-impaired population. The use of

the noise condition alone, although shown to increase the

complexity of the task, does not appear at this time to be a

useful test to separate the centrally impaired from the

peripherally impaired. Further research is needed in that

area. The use of both the conditions together, however, has

been shown to delineate the groups. Lastly, there appears

to exist a small percentage of the older population that

experience both a PI and a CI, yet based on the P3 component

would seem to function as centrally impaired. This issue of

tneM ga npalsmnambte ieistijEgadfitar.

Limitations of the Study

Thirty-one males from the Gainesville Veterans

Administration Medical Center served as subjects for this

investigation of the use of the event-related potential, the

P3 component, as a measure of central auditory function in

an older, hearing-impaired adult population. The subjects

were carefully controlled for age, hearing sensitivity for

500 and 1000 Hz, general health, and cognitive intactness.

Any subject not meeting any of the criteria was dismissed.

Yet despite the carefully controlled low frequency hearing

sensitivity a statistically significant difference in

hearing sensitivity between the groups existed for the high

frequency. The CI group had poorer auditory sensitivity at

2000 and 4000 Hz. Despite controlling for the test

frequencies, the difference in the high frequencies may have

had an influence. Future studies should attempt to match

the audiograms for equivalent hearing sensitive between

groups to control for that possibility.

While it was necessary to have carefully controlled

extraneous variables, it limited the availability of the

population. The sample size in this study was small

considering the amount of variability that existed,

especially for the P3 component. The use of repeated

measures increases the power of the findings, yet the

variability is an area of concern and limitation.

The use of veterans is also a limitation. Over 100

potential subjects were contacted; only 44 agreed to

participate. The men that agreed to participate seemed to

have a very positive attitude toward the Gainesville

Veterans Administration Medical Center. They expressed a

desire to help all that they could since the VA had been so

helpful to them. This willing attitude to participate may

have resulted in a more cooperative nature than might be

expected from other less willing veterans or subjects.

In addition to the limitations imposed by the subject

selection, there are several limitations set by the design

of the study. First, the subjects were classified as a

function of the SSI, so essentially the results of the P3

component are confirming the results of the SSI. To test

whether the P3 is assessing something different from the SSI

would involve the use of another central auditory measure as

the classification tool and statistically test the

differences between the SSI and the P3 component. To date,

there are a number of central tests available, however, the

most widely used is the SSI. The use of another measure,

perhaps a communication inventory instead of a diagnostic

tool or the use of more medical diagnostic tools such as the

CT scan or magnetic resonator (MRI), may be more

appropriate for future research.

Another limitation involves the stimuli. The use of

the 500 and 1000 Hz tones may have been too easy. Perhaps

the use of a finer discrimination may have been more

sensitive, such as 500 and 700 Hz, and may have reduced the


The limitations, while minor, reduce the generaliz-

ations of the findings to the general population. Further

studies need to be conducted with different age groups,

behavioral tests, and P3 stimuli. The P3 component is a

fascinating component that has begun to interest

audiologists and it merits thorough investigation.

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.7 - mvs