Impact of stimulus-related and listener

Impact of stimulus-related and listener-related factors on
cognitive processing load as indicated by pupil dilation
1,4
Barbara Ohlenforst,
1,2,3
Adriana Zekveld,
3,4,5
Thomas Lunner,
4,6
Dorothea Wendt,
7
Graham Naylor,
1,4
Yang Wang,
8
Artur Lorens and Sophia Kramer
1
1) Section Ear & Hearing, Dept. of Otolaryngology-Head and Neck Surgery, VU University Medical Center and EMGO Institute for Health Care Research, Amsterdam; 2) Dep. of Learning & Behaviour, Linköping University, Sweden;
3) Linnaeus Centre HEAD, The Swedish Institute for Disability Research, Linköping, Sweden; 4) Eriksholm Research Centre, Denmark; 5) Department of Clinical and Experimental Medicine, Linköping, Sweden; 6) Technical
University of Denmark, Department of Electrical Engineering, Lyngby, Denmark; 7) Institute of Hearing Research Scottish Section, Glasgow, United Kingdom; 8) Institute of Physiology and Pathology of Hearing, International Center
of Hearing and Speech , Warsaw ,Poland
Background motivation & hypothesis
Different peak pupil dilation (PPD) patterns across intelligibility levels were
previously found between hearing-impaired and normal-hearing listeners [1,4]. The
mean PPD responses increased with decreasing signal-to-noise ratios (SNRs),
resulting in relatively small mean PPDs at low intelligibility levels compared to
conditions of high intelligibility [4]. However, at low and high SNRs it can be
assumed that listening effort is lower either due to ‘giving up’ or due to ‘very easy’
conditions. We assume therefore an inverted u-shape across SNRs for hearingimpaired and normal-hearing listeners.
When speech was masked by a single-talker compared to stationary masker,
larger PPDs were found on average [5]. Based on those previous findings, on
average higher PPD values are expected for the single-talker compared to the
stationary masker.
The aim of the present study was to investigate the PPD for:
Q1: Hearing-impaired versus normal-hearing participants across a wider range of
SNRs
Q2: two different masker types
Method
Average hearing thresholds
Participants
-20
• Mean age for hearing-impaired and
for normal-hearing participants: 47
years
• Criteria
for
normal-hearing
thresholds <= 20 dB HL (500 Hz to 4
kHz) pure-tone average (PTA).
• Hearing-impaired participants had
symmetrical, sensorineural, mild to
moderate hearing loss.
0
HEARING THRESHOLD [DB HL]
• 25 hearing-impaired and 30 normalhearing participants
normal-hearing
hearing-impaired
Mixed-design ANOVA to estimate the effect of different SNRs for the singletalker masker and hearing-status (PTA) on the PPD.
Significant main effect of SNR: F(8,396)=10.30, p=0.00
Significant interaction between SNR and PTA: F(8,396)=6.25, p=0.00
**
**
* p=0.01-0.0.5
** p<0.01
**
Mixed-design ANOVA to estimate the effect of different masker types and
hearing-status (PTA) on the peak pupil dilation (PPD).
Significant main effect of masker type: F(2,939)=4.73, p=0.01.
The PPDs for both masker types differ significantly from sentences in quiet but
they don’t differ from each other.
20
• Averaged PPD responses for stationary noise and interfering talker background
40
60
80
100
500
Stimuli
1000
2000
FREQUENCY [Hz]
4000
• Versfeld sentences [2] in quiet
• Sentences in stationary masker (SNRs:-12 to +16 dB)
• Sentences in single-talker masker (SNRs: -25 to +15 dB)
Individual frequency shaping was applied for the hearing-impaired participants
according to the NAL prescription [3].
Outcome measures:
• Averaged intelligibility scores for stationary noise and interfering talker
background
The PPD relative to the baseline was recorded during stimuli presentation and
sentence recall. Each stimulus consisted of 3 seconds pre-noise, a sentence in
noise (around 2 seconds), and 4 second post-noise.
Speech intelligibility was measured for all conditions.
Analysis and results
Mixed-design ANOVA to estimate the effect of different SNRs for stationary
masker and hearing-status (PTA) on the PPD
Significant main effect of SNR: F(7,356)=7.81, p=0.00
Significant interaction between SNR and PTA: F(7,356)=3.85, p=0.00
Discussion & conclusion
*
* p=0.01-0.0.5
** p<0.01
**
*
References: [1] Kramer et al. 1997; [2] Versfeld et al. 2007; [3]
Byrne & Dillon, 1986; [4] Zekveld et al. 2011; [5] Koelewijn et al.
2014;
Q1: There was a significant interaction between SNRs and hearing status for
both masker types. The expected inverted u-shape across SNRs was
shown.
Q2: There was a significant main effect of masker type on the PPD. The
PPDs for both, the single-talker and the stationary masker differed
significantly from the PPDs for sentences in quiet. Previous findings,
showing larger PPDs for the single-talker compared to the stationary
masker, were not found.
LISTEN607373