Frequency discrimination in children: Perception, learning and attention

Abstract

It is generally believed that both sensory immaturity and inattention contribute to the poor listening of some children. However, the relative contribution of each factor, within and between individuals, and the nature of the inattention are poorly understood. In three experiments we examined the threshold and response variability of 6–11 y.o. children on pure tone frequency discrimination (FD) tasks. We first confirmed that younger children had both higher thresholds and greater within- and between-listener variability than older children and adults. Higher thresholds were mostly attributed to high response variability due to poor sustained attention. We next compared performance on the auditory FD task with that on visual spatial FD. No correlation was found between the thresholds or variability of individuals on the two tasks, suggesting involvement of modality-specific attention. Finally, we found lower thresholds for 8–9 y.o. children performing auditory FD training in a classroom than in the laboratory, possibly due to training session length or to a more familiar, motivating and focussed training environment. The adult-like performance of many younger children at times during their testing or training, together with the high response variability of immature performers, suggested that most elevated FD thresholds in children are due to inattention.

Introduction

Children are generally considered to have poorer attention than adults, but attention is a construct that must be inferred indirectly from variations in performance on specific tasks (e.g. Manly et al., 2001). Psychoacoustic studies have long recognised the intervening effect of inattention on the assessment of hearing in children. In an attempt to separate auditory performance from attention, the properties of psychometric functions relating performance to stimulus level have been measured and modelled. These properties include the slope of the function (Allen and Wightman, 1994) and the extent to which performance at high stimulus levels falls short of perfection (Bargones et al., 1995). In more recent work, auditory spectral distractors (Stellmack et al., 1997, Oh et al., 2001) and informational masking approaches (Wightman and Kistler, 2005) have been used to show immature selective attention in children. Generic, non-psychoacoustic measures of attention, such as the TEA-Ch (Manly et al., 2001), have also revealed immature auditory attention in subtests aimed at specific attentional functions (e.g. sustained, selective and executive control). Psychoacoustic approaches have assumed that performance varies randomly (stochastically) over time; that the psychometric function is a ‘snapshot’ of both perception and attention. The generic approach assumes that attention is essentially a singular and multimodal function; that inattention will be simultaneously and rather indiscriminately manifest in a variety of tasks.

In this paper, we take several novel perspectives on the relation between attention and listening in children. We first examine how children’s auditory performance changes over time. Our premise is that attention is constantly varying, both within and between tasks. The degree to which inattention is contributing to listening should be apparent as short- or medium-term changes in auditory abilities. Children who are attentive should, in a standard staircase adaptive procedure (Amitay et al., 2006), produce a consistent pattern of ‘reversals’ – levels of the stimulus that dynamically mark the upper and lower limits of threshold variability. Here, the staircase would be expected gradually to converge on a threshold, with little fluctuation around that point. Successive threshold determination ‘tracks’ should produce consistent estimates. Inattention, on the other hand, would be expected to lead to a greater degree of performance variability. This should manifest as higher thresholds and a greater range of reversals and inter-track differences. We might expect this to be particularly marked when measuring thresholds in noisy and/or otherwise distracting environments, such as a school classroom.

Leading models suggest that attention has both general (supra- or multi-modal) and modality-specific components (Spence, 2001). In a second experiment, we examine the relation between auditory and multimodal attention and children’s listening by comparing individual performance on closely matched auditory and visual tasks. Previous work (Hawkey et al., 2004) used a similar technique to separate procedural and perceptual aspects of learning. Here, we reason that, if multimodal attention plays a significant role in the hypothesised attention drift described above, there should be a correlation between the auditory and visual tasks on measures of both performance and response variability. A lack of such a relation provides evidence for dominant unimodal influences.

Many studies have shown that auditory training improves performance on a variety of listening tasks, both in adults and children (Merzenich et al., 1996, Moore et al., 2005, Moore and Amitay, 2007, Wright and Zhang, 2006). We have suggested that attention makes a major contribution to such auditory learning by showing, for example, that training on a non-auditory task (the visuospatial computer game, Tetris^®) can improve performance on an auditory frequency discrimination (FD) task (Amitay et al., 2006). Here we report and compare results of training children on FD tasks in the lab (Halliday et al., 2007) and in a school environment. Because of the possible interfering effect of noise on attention, we predicted that the school environment would be less conducive to learning than the quiet of a laboratory sound chamber. We also present data on the variability in children’s performance on the trained task that remains a major challenge for the understanding and control of children’s attention in relation to auditory learning.

The studies reported here thus had three specific aims. First, to examine the influence of attention on children’s FD listening by examining the time course of performance variation. Second, to examine whether changes in children’s listening over short time periods are influenced by multimodal or unimodal mechanisms. And third, to examine whether FD training in children is influenced by the environment in which the training takes place.