Abnormal neural encoding of repeated speech stimuli in noise in children with learning problems

Abstract

Objectives: This study investigated whether neurophysiologic responses to repeated speech stimuli, presented in quiet and noise, differed between normal children (NL) and children with learning problems (LP).

Methods: Subjects were normal-hearing, school-age children. NL subjects scored significantly better than LP subjects on measures of reading, spelling and speech sound discrimination. Stimuli (40 ms /da/) were presented to the right ear at 80 dB SPL. Stimuli were presented in trains of four, separated within trains by 360 ms. The interval between trains was 1060 ms. Stimuli were presented in quiet and in white noise (S/N+15). Cortical responses were recorded from an electrode placed along the midline at Cz.

Results: Correlations between the first and 4th responses were lower in noise than in quiet for LP subjects only. Response correlations in quiet were no different between groups. There were no root-mean-square (RMS) amplitude differences between groups.

Conclusions: Response correlation in noise suggested that the LP population consisted of two subgroups, one whose responses appeared relatively normal, and another whose responses were severely degraded by repetition in noise. Response correlations in noise were related to behavioral measures of auditory processing and spelling. These findings suggest that abnormal, asynchronous, auditory cortical encoding may underlie some language-based learning problems.

Introduction

The ability of a listener to understand real speech in a natural environment places many demands on the auditory system. Among these demands are the accurate representation of rapidly changing spectral information comprising the speech signal and the separation of this signal from background noise. While all listeners clearly demonstrate impaired perception at extremes in speaking rate and background noise, a subset of listeners experiences enhanced sensitivity to the detrimental effects of repeated stimulation and background noise. Many studies have shown that subjects diagnosed with language-based learning disabilities perform poorly when processing rapid acoustic signals (Tallal and Piercy, 1974, Farmer and Klein, 1995, Hari and Kiesila, 1996, Wright et al., 1997, Nagarajan et al., 1999, Cestnick and Jerger, 2000, Temple et al., 2000). Similarly diagnosed subjects have exceptional difficulty processing acoustic signals which are presented in the presence of noise (Jerger et al., 1987, Breedin et al., 1989, Katz, 1992, Katz et al., 1992, Welsh et al., 1996, Bellis, 1996, Chermak and Musiek, 1997, Cunningham et al., 2001). Neurobiological abnormalities accompany many of these auditory processing deficits (Nagarajan et al., 1999, Temple et al., 2000, Cunningham et al., 2001). These findings contribute to an understanding that some language-based learning disabilities are rooted, in part, in altered representations of acoustic information. Distorted encoding of acoustic speech signals could underlie weakened perception and categorization of phonemic information (Kraus et al., 1996, King et al., 2002). Such deficits could certainly contribute to difficulties in the development of reading and other language skills (Godfrey et al., 1981, Reed, 1989, McBride-Chang, 1996).

While the primary goal of the current study was to understand auditory processing in children with learning problems, there was also motivation to further understand normal neural mechanisms which underlie processing of repeated stimuli, and stimuli in noise, especially when these stresses are combined, as they are in most real listening environments. Studies of normal subjects have shown that cortical responses decrease in amplitude and increase in latency upon repetition of stimuli (Woods and Elmasian, 1986, Budd et al., 1998) and following the addition of noise (Whiting et al., 1998, Martin et al., 1999).

Inspired by the aforementioned findings, the present study was designed to further examine effects of stimulus repetition and background noise on the neural representation of auditory stimuli in normal and learning impaired subjects. Specifically, the intent was to expand upon previous studies, which investigated rapid or repetitive stimuli or noise in isolation, often using simple tonal stimuli. By simultaneously stressing the auditory system with stimulus repetition and background noise, and by using a complex speech stimulus, the present study incorporated conditions which simulated real listening situations more accurately than studies which incorporated some of these features in isolation. Additionally, non-linear transformations throughout the auditory system make difficult the prediction of responses to complex signal patterns based on knowledge of responses to simpler stimuli (Sachs et al., 1983, Rauschecker, 1997). Responses elicited by our paradigm thus provide further insight into the neural representation of speech under ‘real world’ conditions. Since learning disabilities manifest themselves outside of the laboratory in normal, everyday life, it is by mimicking real conditions that we may most accurately describe any neural abnormalities underlying such problems.