1. Neuroscience
Download icon

Individual differences in selective attention predict speech identification at a cocktail party

  1. Daniel Oberfeld  Is a corresponding author
  2. Felicitas Klöckner-Nowotny
  1. Johannes Gutenberg-Universität, Germany
Research Article
Cite this article as: eLife 2016;5:e16747 doi: 10.7554/eLife.16747
7 figures and 3 tables

Figures

Average audiometric hearing thresholds (in dB HL), at octave frequencies between 125 Hz and 4 kHz (N = 50).

Left panel: left ear. Right panel: right ear. Error bars represent 95% confidence intervals.

https://doi.org/10.7554/eLife.16747.005
Average individual proportion correct (speech recognition score; SRS) in the simulated cocktail-party listening task with two spatially separated interfering speakers (N = 50).

This measure served as the criterion variable in the regression analyses. The mean (M) and the standard deviation (SD) are displayed.

https://doi.org/10.7554/eLife.16747.006
Schematic depiction of the two-interval intensity discrimination task used to measure auditory selective attention.

Green: target tones. Red: backward maskers ('distractors'). The standard (S) was a 1 kHz tone presented at 60 dB SPL. An intensity increment (I) was presented in either the first or the second interval, with equal a-priori probability. The task was to select the interval containing the louder target (that is, standard-plus-increment). The maskers were 1 kHz tones presented at 90 dB SPL. The same temporal configuration was used in the in-quiet condition, except that the maskers were not presented.

https://doi.org/10.7554/eLife.16747.007
Individual elevation of the intensity difference limen caused by the backward maskers (DLelev), defined as the difference between the DL under masking and the DL in quiet.

Lower values represent a better capability of directing auditory selective attention to the target tones. N = 50.

https://doi.org/10.7554/eLife.16747.008
Individual IPD thresholds in the TFS-LF test (TFSth).

Lower values represent better binaural sensitivity for the temporal fine structure. N = 50.

https://doi.org/10.7554/eLife.16747.009
Example stimuli from the flanker task (Kramer and Jacobson, 1991) used to measure spatial visual selective attention.

The participants' task was to decide whether the central target line was dotted or dashed. In the neutral condition (left column), the neighboring flanker lines were solid and therefore not associated with a response. In the incompatible condition (right column), the flanker lines were associated with the incompatible response. The horizontal lines and the colors promoted the perceptual grouping of the target line and the flankers as either belonging to the same object (upper row), or into different objects (lower row).

https://doi.org/10.7554/eLife.16747.010
Individual flanker interference (IntFlanker) in the visual attention task.

Lower values represent a better capability of directing visual selective attention to the target line. N = 50.

https://doi.org/10.7554/eLife.16747.011

Tables

Table 1

Results of the multiple regression analysis. Criterion variable: speech recognition score (SRS; proportion correct) in the simulated cocktail-party listening task. Predictors: age, masker-induced elevation of the intensity difference limen (DLelev), the amount of flanker interference in the flanker task (IntFlanker), IPD threshold in the TFS-LF task (TFSth), pure-tone average thresholds on the better ear (PTABE), average asymmetry in the hearing thresholds between left and right ear (HLdiff), intensity-DL in quiet (DLquiet), response time in the neutral condition of the flanker task (RTneutral), and proportion of correctly recalled consonants in the working memory task (SSPcorr). All variables were z-standardized.

https://doi.org/10.7554/eLife.16747.002
PredictorβSEtpGDWβLasso
Intercept0.0680.0960.7100.4800.081
Age0.1940.1101.7600.0860.020
DLelev−0.3470.1073.2400.0030.152−0.220
IntFlanker−0.2330.1032.2700.0290.052−0.081
TFSth−0.3830.1033.7300.0010.204−0.286
PTABE0.1370.1021.3500.1860.016
HLdiff−0.0880.1060.8300.4130.007
DLquiet−0.0070.1140.0700.9480.021
RTneutral−0.0370.1290.2800.7780.015
SSPcorr0.1930.1111.7400.0910.0850.089
R2 =0.57
p<0.001
R2 =0.44
  1. Note: N = 45. β: estimated ordinary least-squares (OLS) regression coefficient. SE: standard error of the estimate. t: t-statistic. Bold font indicates a β significantly different from 0 (p<0.05). GDW: general dominance weight. βLasso: regression coefficients for predictors selected by the Lasso procedure (model selection via four-fold cross-validation).

Table 2

Pairwise Pearson partial correlation coefficients, controlling for age. N = 50. In each row, the upper numbers are the partial correlation coefficients (ρpartial), and the lower numbers are the p-values for the test of |ρpartial| > 0. The rightmost column shows Pearson correlation coefficients with age. Bold font: p<0.05. Italics: p<0.10.

https://doi.org/10.7554/eLife.16747.003
DLelevIntFlankerTFSthPTABEHLdiffDLquietRTneutralSSPcorrSSQspeechSSQspatialSSQqualitiesAge
OLSAPcorr−0.374−0.149−0.3530.060−0.163−0.244−0.2320.3380.1210.0830.2300.033
0.0080.3070.0130.6830.2630.0910.1090.0180.4070.5700.1120.819
DLelev−0.0470.0380.030−0.083−0.0450.092−0.255−0.302−0.049−0.178−0.045
0.7480.7930.8380.5700.7590.5290.0770.0350.7400.2200.754
IntFlanker−0.081−0.018−0.234−0.032−0.094−0.112−0.141−0.098−0.1120.045
0.5780.9030.1050.8260.5220.4440.3350.5010.4440.758
TFSth0.034−0.0230.3990.312−0.177−0.149−0.314−0.3520.027
0.8180.8730.0050.0290.2240.3060.0280.0130.852
PTABE−0.292−0.083−0.010−0.0920.1360.0820.097−0.248
0.0420.5720.9440.5310.3530.5770.5080.082
HLdiff0.1950.0380.0870.2740.1980.1190.227
0.1800.7950.5510.0570.1740.4160.113
DLquiet0.383−0.115−0.008−0.118−0.1300.082
0.0070.4310.9590.4200.3750.573
RTneutral−0.198−0.087−0.135−0.0110.217
0.1720.5520.3550.9400.129
SSPcorr−0.0110.0810.053−0.380
0.9400.5800.7200.006
SSQspeech0.7070.7280.040
<0.0001<0.00010.784
SSQspatial0.7010.123
<0.00010.393
SSQqualities0.058
0.692
Table 3

Multiple regression analysis of the relation between the SSQ scores (Gatehouse and Noble, 2004) representing self-reported hearing abilities (predictors) and the speech recognition score in the simulated cocktail-party listening task (criterion). N = 47.

https://doi.org/10.7554/eLife.16747.004
PredictorβSEtp
Intercept−1.1500.9881.1600.251
Age−0.0060.0340.1700.864
SSQspeech0.0060.1160.0500.958
SSQspatial−0.1600.1171.3600.181
SSQqualities0.3210.1422.2500.030

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)