Development and the role of internal noise in detection and discrimination thresholds with narrow band stimuli

Abstract

The experiments reported here examine the role of internal noise in the detection of a tone in narrow band noise and intensity discrimination for narrow band stimuli in school-aged children as compared to adults. Experiment 1 used 20-Hz wide bands of Gaussian and low-fluctuation noise centered at 500 Hz to assess the role of stimulus fluctuation in detection of a 500-Hz pure tone. Additional conditions tested whether performance was based on level and/or level-independent cues. Children's thresholds were elevated with respect to adults, and whereas adults benefited from the reduced fluctuation of low-fluctuation noise, children did not. Results from both groups were consistent with the use of a level cue. Experiment 2 estimated intensity increment thresholds for a narrow band Gaussian noise or a pure tone, either with or without a presentation-by-presentation level rove, an additional source of level variability. Stimulus variability was found to have a larger effect on performance of adults as compared to children, a rather counterintuitive finding if one thinks of children as more prone to informational masking introduced by stimulus variability. Both tone-in-noise and intensity discrimination data were consistent with the hypothesis that children's performance is limited by greater levels of internal noise.

FIG. 1

Example stimuli from Experiment 1 are shown, with Gaussian noise (left panel) and low-fluctuation noise (right panel), and signal conditions indicated in the left margin. The top row in each panel shows the segments of the stimulus identified as the fringe and core portions. The next row, labeled all cues, shows a standard stimulus (in gray) and the envelope of that standard summed with a 65-dB SPL pure tone signal (black line). The final two rows show standard and signal-plus-standard for the comparable level-only and no-level conditions. For illustrative purposes a single random sample was selected for all three signal conditions within each panel; in the experiment a new random sample was generated prior to each presentation.

FIG. 2

The mean thresholds from Experiment 1 are plotted for Gaussian noise (top panel) and low-fluctuation noise (bottom panel). Bar markings indicate mean data from adults (□) and children (⊠), and the circles associated with each bar show individual data. Stimulus conditions are indicated along the abscissa. The ratios at the right of each panel indicate the proportion of listeners contributing to each mean in the no-level conditions: individuals’ data were omitted from the average if they were deemed to be indistinguishable from ceiling performance.

FIG. 3

Individual data from Experiment 2 are shown in units of dB as if the signal had been added in random phase. The ordinate labels at the right show associated units of ΔL. Panels show results for adults (left panel) and children (right panel), with stimulus conditions indicated along the abscissa. Dark lines indicate fits to the data.

FIG. 4

Individual data of child listeners in the tone-steady condition are plotted as a function of age. Consistent with units used to compute internal noise, data are plotted in units of ΔL, with ±2 std around the mean adult threshold indicated as the shaded region.