Key Words
aging - cognition - hearing in noise - P300
INTRODUCTION
Older adults (OAs) often report difficulty hearing in background noise which is not
completely attributable to peripheral hearing loss ([Dubno et al, 1984]; [Gordon-Salant, 1987]; [Stuart and Phillips, 1996]; [Gordon-Salant and Fitzgibbons, 1997]; [Snell et al, 2002]; [Pichora-Fuller, 2003]). Researchers have investigated other age-related changes that might impact the
ability to hear in background noise. Investigators have demonstrated that hearing
in background noise is a complex process that necessitates the use of multiple systems,
including cognition ([Pichora-Fuller et al, 1995]; [Tun et al, 2002]; [Anderson et al, 2013]; [Helfer and Freyman, 2014]). Although age-related declines in cognition and hearing in background noise occur,
the underlying age-related changes in the processing of auditory stimuli in background
noise are yet to be fully understood.
Auditory cortical potentials, specifically N1 and P2, have been used to objectively
document changes in auditory cortical processing in background noise in younger adults
(YAs) and OAs. The N1 and P2 potentials have long been reported to be exogenous potentials
with neural generators in the auditory cortices ([Vaughan and Ritter, 1970]; [Elberling et al, 1982]; [Scherg and Von Cramon, 1986]; [Scherg et al, 1989]). Researchers have reported significantly longer N1 and P2 latencies when comparing
quiet with noise conditions in YAs ([Billings et al, 2009]; [McCullagh et al, 2012]) and OAs ([McCullagh and Shinn, 2013]; [Billings et al, 2015]) as well as significantly longer P2 latencies for OAs compared with YAs ([McCullagh and Shinn, 2013]). Significantly reduced amplitudes of N1 ([Billings et al, 2009]) and P2 ([Billings et al, 2009]; [McCullagh et al, 2012]) in noise compared with quiet conditions in YA listeners have also been demonstrated.
Significantly larger N1 and P2 amplitudes were found for OAs compared with YAs in
the quiet condition ([McCullagh and Shinn, 2013]; [Billings et al, 2015]). Conflicting evidence exists regarding differences in noise between YAs and OAs.
Although [Billings and colleagues (2015)] demonstrated significant amplitude differences in noise when comparing OAs and YAs,
others have found no significant differences between groups in the poorest noise condition
tested [0 signal-to-noise ratio (SNR)] ([Kim et al, 2012]; [McCullagh and Shinn, 2013]). Methodological variations likely led to the differences in findings across studies;
however considered together, these findings suggest that auditory cortical responses
vary in noise, at least to some degree, in YAs as compared to OAs.
Converse to the auditory N1 and P2, the auditory P300 has long been known to be an
endogenous potential and has the ability to establish the cognitive aspects of processing
auditory stimuli ([Sutton et al, 1965]; [Goodin et al, 1983]; [Squires and Hecox, 1983]; [Kibbe-Michal et al, 1986]). It should be noted, however, evidence exists to support contributions that are
exogenous in nature as well ([Stapleton and Halgren, 1987]; [Stapleton et al, 1987]; [McCullagh et al, 2009]). The P300 is elicited using an oddball paradigm, and the listener must make a decision
regarding the stimuli to produce a response ([Sutton et al, 1965]; [Stapleton et al, 1987]), and P300 latencies are extended as the complexity of the task becomes more challenging
([Polich, 1987]; [Gaál et al, 2007]). Findings suggest that as task complexity increases, a delay in cognitive processing
of the auditory stimuli occurs. These findings have implications for the processing
of more complex auditory stimuli in the presence of noise.
Investigators have researched the complex task of listening in background noise using
the auditory P300 in adult listeners. They have demonstrated significantly longer
P300 latencies ([Polich et al, 1985]; [Obert and Cranford, 1990]; [Salisbury et al, 2002]; [McCullagh et al, 2012]) and reduced P300 amplitudes ([Obert and Cranford, 1990]) when background noise was present in adults. Although [McCullagh et al (2012)] included only YA listeners in their study, both [Obert and Cranford (1990)] and [Salisbury et al (2002)] included young and middle-aged adults in their investigations. They consistently
demonstrated delays in cognitive processing of auditory stimuli in background noise
in adult listeners; however, age-related changes in processing were not specifically
addressed in these studies.
Age-related changes in P300 in quiet conditions have been observed ([Goodin et al, 1978]; [Pfefferbaum et al, 1980]; [1984]; [Picton et al, 1984]; [Anderer et al, 1996]). Significantly longer P300 latencies and reduced P300 amplitudes in OA listeners
as compared to YA listeners were reported ([Goodin et al, 1978]; [Picton, et al, 1984]; [Anderer et al, 1996]). Conversely, others did not find significant reductions in P3 amplitude between
OAs and YAs ([Pfefferbaum et al, 1980]). Differences between these outcomes are likely related to age-related hearing loss,
stimulus presentation levels, and recording electrode site differences between studies.
These results do, however, suggest some age-related declines in synchronous firing
and rate of transmission of the auditory cortical neurons contributing to the P300
potential (reviewed by [Brody et al, 1975]; [Goodin et al, 1978]).
Although age-related changes in the auditory P300 have been reported in quiet, little
data in the literature exist regarding the age-related changes in the P300 in background
noise. Although no significant age-related changes in P300 latency in contralateral
noise were reported ([Cranford and Martin, 1991]; [Bertoli et al, 2005]), some researchers have reported significant age effects for P300 amplitude in contralateral
noise ([Bertoli et al, 2005]), whereas others have not ([Cranford and Martin, 1991]). These two studies were limited by the utilization of monaurally presented stimuli
in contralateral noise only. Because hearing in noise is typically a binaural process,
the binaurally recorded P300 allows for a better determination of the age-related
changes in cognitive processing of auditory stimuli in background noise. Furthermore,
the methodologies of these studies are limited by the utilization of one SNR. Eliciting
the P300 at multiple SNRs allows for the comparison of the changes that occur in cognitive
processing of the auditory stimuli as the task complexity increases (i.e., decreasing
SNR). Given that conflicting evidence exists related to the changes in cognitive processing
of auditory stimuli and a lack of evidence exists related to binaurally-recorded P300
evoked responses at multiple SNRs, additional research is warranted. Therefore, the
purpose of this study was to investigate age-related differences in cognitive processing
of auditory stimuli by evoking the auditory P300 at multiple SNRs.
METHODOLOGY
Participants
Thirty-five adults were recruited from the University and neighboring communities
to participate in this study. All participants reported negative otologic and neurological
histories. Otoscopy was within normal limits bilaterally for all participants. Participants
were divided into a YA group (n = 20; mean age = 21.1 yr, standard deviation = 2.7
yr) and an OA group (n = 15; mean age 66.4 yr, standard deviation = 4.6 yr). All OA
participants were administered a backward digit span task to assess working memory
and had to perform within normal limits ([Gregoire and Van der Linden, 1997]) to be included in the study. Data from the normal-hearing YA group were previously
reported in [McCullagh et al (2012)]. The Southern Connecticut State University Institutional Review Board approved this
study, and all participants signed an informed consent form before testing.
Hearing sensitivity for all groups was established using the modified Hughson-Westlake
procedure ([Carhart and Jerger, 1959]). All participants presented with no air-bone gaps greater than 10 dB HL and no
interaural asymmetries greater than 15 dB HL. All YA participants had pure tone thresholds
better than 25 dB HL for the octave frequencies of 250 to 4000 Hz and for 6000 Hz,
bilaterally. All OA participants had normal hearing sensitivity for the octave frequencies
of 250 to 1000 Hz and thresholds no poorer than 55 dB HL at 2000, 4000, and 6000 Hz,
bilaterally. All participants had normal hearing sensitivity at the two frequencies
used in the electrophysiological procedure (500 and 1000 Hz). Mean audiometric data
for the YA and OA groups are presented in [Table 1]. Mean thresholds were within 10 dB HL at 500 and 1000 Hz between groups. Results
from independent t tests demonstrated no statistically significant differences between the two groups
for the left ear 500 Hz thresholds (t = −0.18, p = 0.86). Results did demonstrate statistically significantly differences between
the two groups for the thresholds for the following: left ear at 1000 Hz (t = −4.38, p < 0.001); right ear at 500 Hz (t = −5.25, p < 0.001); right ear at 1000 Hz (t = −3.73, p < 0.001).
Table 1
Audiometric Data Including Mean 500- and 1000-Hz Thresholds, three-Frequency Pure
Tone Averages (PTA) and High Frequency Pure Tone Averages (HFPTA) for the YA and OA
Groups
|
Left Ear
|
Right Ear
|
|
500-Hz
|
1000-Hz
|
PTA
|
HFPTA
|
500-Hz
|
1000-Hz
|
PTA
|
HFPTA
|
|
YAs
|
5.67
|
4.00
|
4.56
|
4.67
|
7.33
|
4.00
|
4.33
|
4.22
|
|
OAs
|
12.33
|
11.33
|
13.11
|
17.11
|
13.67
|
12.33
|
15.11
|
19.33
|
Procedures
All participants were seated in a double-walled sound treated booth. N1, P2, and P300
were measured using an Intelligent Hearing Systems (IHS) SmartEP system with silver
chloride cup electrodes affixed to Cz (noninverting) and the nape of the neck (inverting).
Ground and eye-blink electrodes were affixed to the high forehead and outer canthus
of the eye, respectively. Impedances were below 3 k ohms and balanced across electrode
sites. Only one noninverting electrode (Cz) was used to establish a protocol that
was considered clinically feasible. Neuroelectric activity was filtered from 1 to
30 Hz with a 12 dB/octave roll-off. A standard oddball paradigm (80% frequent tones;
20% target tones) was used to evoke the potentials. Forty millisecond (10 msec rise/fall;
20 msec plateau) 500-Hz stimuli (frequent) and 1000-Hz (target) stimuli were presented
binaurally at a repetition rate of 0.9 per sec. Responses were obtained binaurally
for the purposes of ecological validity and to replicate “real-world” listening environment.
Participants were asked to count the number of 1000-Hz (target) tones presented for
each trial. All waveforms were replicated and an average of the two waveforms (approximately
600 trials total; 300 original and 300 replicated) was used for amplitude and latency
measurements. The number of rejected trials was maintained at less than 10% for inclusion
of data. N1 and P2 data were collected and those results were previously reported
in [McCullagh and Shinn (2013)].
The P300 waveforms were acquired in four listening conditions (quiet, +20, +10, and
0 SNR). Listening condition order was counter-balanced across participants. Binaural
behavioral thresholds were established to the 500- and 1000-Hz stimuli generated by
the Intelligent Hearing Systems (IHS) SmartEP system as well as the white noise generated
by the Grason Stadler Inc (GSI) 61 diagnostic audiometer to establish the different
SNRs. The 500- and 1000-Hz stimuli were routed through ER-3A insert earphones with
open-fit hearing aid domes affixed on the ends to keep the ear canals free from occlusion
([Weihing and Musiek, 2008]). For the noise conditions, continuous white noise was presented through a speaker
positioned at zero degrees azimuth 42″ from the participants’ nose. The 500- and 1000-Hz
stimuli were presented at 50 dB SL re: behavioral tone burst threshold, and the SNRs
in the noise conditions were achieved by altering the white noise intensity levels
([McCullagh and Shinn, 2013]). Before initiation of the protocol, levels were calibrated using a sound level
meter to ensure accurate SNRs.
P300 amplitudes and latencies were measured for each listening condition (quiet, +20,
+10, and 0 SNR). P300 was considered the most positive peak between 245 and 450 msec
poststimulus onset. When a bifid or broad peak was present, the midpoint of the waveform
was used as the latency measurement. P300 amplitude was measured peak-to-trough from
the highest peak of P300 to the closest negative trough following P300 ([Jirsa, 1992]; [McCullagh et al, 2012]). Waveforms were analyzed by two experienced researchers with extensive experience
in auditory late potentials. One of the two reviewers was blinded to the condition
in which the waveforms were acquired.
Statistical Design
Repeated measures analyses of variance (ANOVA) were conducted for the amplitude and
latency measures of the P300. The analyses were performed using SPSS statistical software
(SPSS, Inc., Release 19.0.0). The analyses included the within-subjects factor of
listening condition (quiet, +20, +10, and 0 SNR) and between-subjects factor of group
(YA and OA). Greenhouse–Geisser corrected degrees of freedom were used when sphericity
could not be assumed ([Greenhouse and Geisser, 1959]).
Post hoc paired t-tests were conducted to determine where significant differences within participants
existed. Paired t-tests for all combinations of listening conditions were conducted. In addition, planned
comparisons were performed for the following conditions: quiet and +20, +20 and +10
SNR, +10 and 0 SNR, and quiet and 0 SNR. Bonferroni adjustments were performed at
the alpha level of 0.05 to adjust for the number of t-tests conducted for each measure.
RESULTS
Descriptive Statistics
Descriptive statistics for P300 latencies and amplitudes for each group at each SNR
(quiet, +20, +10, and 0 SNR) are displayed in [Figures 1] and [2]. Mean P300 latencies increased from the quiet condition to the 0 SNR condition for
each group. Mean P300 latencies were longer for the OA compared with the YA group
in each listening condition. Mean P300 amplitudes decreased for the YA group, but
remained relatively stable for the OA group in each listening condition. Mean P300
amplitudes were between 0.6 and 2.1 μV greater for the YA group in the each of the
listening conditions, with the greatest differences in amplitude occurring in the
quiet condition.
Figure 1 Mean P300 latencies for the older and younger adult groups in quiet and at +20, +10,
and 0 SNRs.
Figure 2 Mean P300 amplitudes for the older and younger adult groups in quiet and at +20,
+10, and 0 SNRs.
Latency
Repeated measures ANOVAs were conducted to determine if significant differences occurred
for P300 latencies in the different listening conditions (quiet, +20, +10, and 0 SNR)
between groups. Results of the repeated measures ANOVAs indicated significant latency
differences between listening conditions [F
(1,3) = 17.03, p < 0.001]. Latencies were significantly longer for the OA compared with the YA groups
[F
(1,3) = 8.48, p = 0.01]; however, no significant group × listening condition interactions existed
[F
(1,3) = 0.65, p = 0.59].
Post hoc comparisons were conducted for the different noise levels to determine where
significant differences existed. The result of the Bonferroni adjustment was an alpha
level of 0.0125. Significant latency differences did not exist between the quiet and
+20 SNR conditions (t = −2.20, p = 0.03), or the +20 and +10 SNR conditions (t = −0.86, p = 0.40). However, significantly longer P300 latencies existed in the 0 SNR condition
compared with the quiet condition (t = −5.99, p < 0.001) and the 0 SNR condition compared with the +10 SNR condition (t = −4.77, p < 0.001).
Amplitude
Repeated measures ANOVAs were conducted to determine if significant differences existed
for the P300 amplitudes in the different listening conditions (quiet, +20, + 10, and
0 SNR) between the two groups. Results of the repeated measures ANOVAs indicated no
significant amplitude differences between listening conditions [F
(1,3) = 0.86, p = 0.46]. Amplitudes were not significantly different for the YA compared with the
OA group [F
(1,3) = 0.79, p = 0.38]. No significant group × listening condition interactions existed [F
(1,3) = 0.50, p = 0.68].
DISCUSSION
The purpose of the study was to ascertain the age-related changes in the auditory
P300 in background noise. P300 responses were elicited in YAs and OAs by binaurally
presenting stimuli in quiet and at three SNRs (+20, +10, and 0 SNR). Results from
this study demonstrated P300 latencies were significantly longer in OAs in noise at
the most challenging condition (0 SNR) compared with the quiet condition and between
the +10 and 0 SNR conditions. Although OAs had significantly longer P300 latencies
compared with YAs, no significant group by listening condition interaction existed.
No significant P300 amplitude differences were found for group, noise, or group ×
listening condition interactions.
Although the effects of peripheral hearing cannot be completely ruled out, attempts
were made to minimize the impact of this potential confounding variable. All participants
in the YA and OA groups had normal peripheral hearing sensitivity for the two test
stimuli (500 and 1000 Hz). The majority (66%) of the OA participants had bilaterally
normal peripheral hearing sensitivity from the octave frequencies of 250 to 4000 Hz
and the interoctave, 6000 Hz. In addition, the test stimuli were presented at 50 dB
SL re: 500 and 1000 Hz thresholds to further minimize the impact of audibility on
the study outcomes.
P300 latencies were significantly longer in noise than in quieter conditions. Longer
P300 latencies were found between the poorer, and likely more auditory demanding,
SNRs (+10 and 0 SNR; quiet and 0 SNR). These findings are similar to previous studies
which demonstrated prolonged P300 latencies in background noise ([Polich et al, 1985]; [Obert and Cranford, 1990]; [Salisbury et al, 2002]; [McCullagh et al, 2012]). Investigators have also previously indicated that P300 latencies increase with
greater task complexity suggesting delayed cognitive processing with more challenging
conditions ([Polich, 1987]; [Gaál et al, 2007]). Because P300 latencies do not significantly increase until more noise is present
(0 SNR), it suggests that cognitive processing of the auditory stimuli remains relatively
stable at easier SNRs, but becomes significantly delayed (poorer) at the most difficult
listening condition (0 SNR).
P300 latencies were significantly longer for OAs than YAs, but a significant group
by listening condition interaction did not exist. Significantly longer P300 latencies
in OAs compared with YAs have been previously reported ([Cranford and Martin, 1991]; [Bertoli et al, 2005]; [Gaál et al, 2007]; [Kropotov et al, 2016]) and indicated an age-related slowing of processing and/or increased inhibition
which contributes to the increases in P300 latency ([Papanicolaou et al, 1984]; [Peters, 2002]; [Gaál et al, 2007]). Because P300 latency has been attributed to stimulus processing rather than response
processing (reviewed by [Rossini et al, 2007]), P300 latency increases in older compared with YAs suggest that OAs take longer
to classify stimuli than YAs. The current study demonstrated the lack of a significant
interaction between age and listening condition on the P300 latency, suggesting that
as the SNR decreases, processing time increases across ages in general. Other researchers
have also failed to report significant interactions between age and P300 latencies
in background noise ([Cranford and Martin, 1991]; [Bertoli et al, 2005]). Thus, support exists for a reduction in stimulus processing speed related to age,
but cognitive processing in background noise is not more delayed in OAs than in YAs.
One unexpected finding was that no statistically significant P300 amplitude differences
for group, noise, or group × listening condition interactions were observed. Lack
of a main effect of noise in adult listeners as well as a lack of an age effect on
P300 amplitude is consistent with previous findings ([Cranford and Martin, 1991]; [Salisbury et al, 2002]). These findings suggest that the amount of neural substrate synchronously responding
did not significantly change in noise compared with quiet conditions and also did
not significantly change because of age. Thus, as task complexity increased with increasing
noise level, cortical neurons contributing to the P300 did not significantly change
for either the YA or OA groups. [Salisbury et al (2002)] explained a lack of P300 amplitude difference in noise to reflect a counteraction
between the increase in resource allocation necessary for increased task complexity
with the reduction in information transfer that occurs in noise. Researchers have
since confirmed the reduction in amplitude of exogenous auditory cortical responses
in noise (N1 and/or P2) ([Billings et al, 2009]; [McCullagh et al, 2012]; [McCullagh and Shinn, 2013]; [Billings et al, 2015]) suggesting the existence of the reduction in information transfer in noise that
[Salisbury et al (2002)] discussed. Moreover, aging studies using fMRI and positron emission tomography to
assess cognitive tasks have demonstrated increased bilateral activation in the brain
([Reuter-Lorenz et al, 1999]; [Cabeza, 2001]; [Charroud et al, 2015]) which may be related to the increase in resource allocation. Researchers have also
suggested that frontal areas, specifically the prefrontal cortex, have increased activation
in cognitive processing tasks and may be a compensatory mechanism in OAs ([Fabiani and Friedman, 1995]; [Fabiani et al, 1998]; [Reuter-Lorenz and Cappell, 2008]; [van Dinteren et al, 2014]). We hypothesize that perhaps compensatory mechanisms may have been activated in
background noise in the OAs, thus reducing the influence of background noise on the
P300 amplitude. Although not elicited in background noise, other researchers have
not found significant age-related P300 amplitude differences with tonal target stimuli
([Katayama and Polich, 1996]; [Gaál et al, 2007]). [Gaál et al (2007)] did not demonstrate a significant interaction between age and task complexity on
P300 amplitude which is similar to the current study. [Katayama and Polich (1996)] did not find significant P300 amplitude differences across differing P300 tasks
(1-, 2-, and 3-tone oddball paradigms) which they explained was an indication that
the P300 is an index of cognitive processing because changing the stimuli did not
change the P300 amplitude.
It should also be noted that more variability in P300 amplitude existed for OAs as
compared with YAs in the quiet condition that was not observed in the noise conditions.
Fjell and Walhovd (2007) demonstrated trial-to-trial variability in P300 amplitudes
that positively correlated with age. It may be the case that there is an inherent
variability in OAs that is difficult to control for when evaluating late auditory
evoked potentials. In addition, sample size may be a contributory factor to the variability
observed in the present investigation. A larger sample size may have reduced the variability
and potentially yielded statistically significant differences in amplitude measurements.
These results suggest that P300 amplitude is likely related to primarily endogenous,
or cognitive, processing and that aging and background noise did not significantly
impact the amplitude; however, exogenous contributions cannot be ruled out. Additional
research is warranted to determine the relationships between endogenous and exogenous
interaction, task complexity, resource allocation, and cognitive processing in YAs
compared with OAs.
CONCLUSION
This present investigation is unique in that age-related changes are studied by eliciting
the auditory P300 binaurally and at multiple SNRs. Results from the present study
provide evidence that auditory cortical processing, regardless of age, is poorer at
more difficult SNRs. However, it also demonstrates that OAs perform significantly
poorer than their YA counterparts. This supports the notion that some degree of age-related
decline in synchronous firing and rate of transmission of the auditory cortical neurons
contributing to the auditory P300 exists. Further studies are needed to elucidate
the impact of noise on auditory cortical processing across populations. Moreover,
although the evaluation of the ability to process in noise is primarily focused on
behavioral measures, objective measures such as those described in this protocol need
to be created to provide additional means by which to evaluate processing in noise
in clinical populations.
Abbreviations
ANOVA:
analyses of variance
OA:
older adult
SNR:
signal-to-noise ratio
YA:
younger adult
