We have proposed a framework for understanding individual difference in the facilitative effects of auditive white noise on performance. As predicted the results show different effects of noise in attentive and inattentive children selected from the normal population. There was significant improvement in performance for the children rated as inattentive by their teachers, and a significant decline in performance for those rated as attentive as noise levels were increased. Furthermore these effects seem independent of other factors measured in the study - attentional ability seems to be a key marker of this effect. Even if inattention and hyperactivity are strongly correlated, no correlation between hyperactivity and a positive noise effect was found, this suggests that in this study inattention is the key factor to explain noise improvement. These results are similar to those previously reported with ADHD patients . Here we discuss theoretical and practical implications of these findings.
From a theoretical point of view the findings are consistent with the suggestion that the neural noise level associated with dopamine tone in inattentive children is sub-optimal, see also  and that noise may enhance performance through the phenomenon of stochastic resonance (SR). According to the model developed here, noise in the environment yields an input to the perceptual system, which can either compensate for low noise in the neural system leading to an output consisting of improved cognitive performance, or, depending on pre-existing levels of neural noise, can add too much to an already well functioning system. The specific neuro-biological and neuro-chemical mechanisms responsible for these effects need further research. For instance, auditive white noise may have its effects either at a perceptual or neuro-psychological level or may operate a neuro-chemical level directly altering levels of dopamine release [13, 51]. Animal models of dopamine function or pharmacological probes to manipulate tonic and phasic dopamine are called for to investigate these effects.
Stimulant medication (e.g. methylphenidate) also improves cognitive performance in children with ADHD [52, 53]. This medication increases dopamine levels by blocking the dopamine transporter . Low performing healthy controls also benefit from increased dopamine transmission, which is manifested in improved cognitive performance and increased prefrontal cortical activity . Our data show that auditory white noise may exert potentially similar effects on cognition as medication through the phenomenon of stochastic resonance (SR). White noise is characterized by randomness and so introduces variability in the nervous system . A poorly tuned neural system benefit from additional white noise. In fact, the stochastic resonance theory predicts that noise that is applied to the signal as an input to a neural cell, improves the signaling efficiency of the output of that cell, where the non-linearity in the firing threshold of the neural cells is the key to improvement of the signal to noise ratio [14, 55].
Despite the fact that it appears that noise and methylphenidate both improve cognitive performance, the underlying mechanism that is the basis of these phenomena are likely to be different. According to the model theoretically speaking the difference between these phenomena is clear. Methylphenidate changes the strength (but not the variability) of the input, which is typically modeled by the gain parameter in abstract neural networks . In contrast, noise changes the variability of the input (but per definition does not influence the strength) of the input. However, despite these clear differences in the underlying level, the behavioral outcome may be similar, and the two mechanisms interact in a complex way, making it difficult to distinguish the phenomena at the behavioral level . Furthermore, direct evidence of difference between these levels is emerging. Pålsson and Söderlund et al. (Noise benefit in pre-pulse inhibition of the acoustic startle reflex, submitted) studied the effect of methylphenidate, dopamine and noise on the startle response in a rat model of ADHD. They found that both control and ADHD strains (SHR) benefited from noise; however, this effect was also found in dopamine lesioned rats, suggesting that dopamine is not a necessary requirement for the stochastic resonance phenomenon to occur.
Another theoretical interpretation of the data is that noise in a general way increase arousal that makes the subject more alert, and less drowsy. The optimal stimulation model states that hyperactivity is as a homeostatic response to underarousal in order to achieve an optimal arousal level . However, this model does not make any explicit predictions about the selective effects of external stimulation whereas in the current study inattentive persons benefited from extra stimulation and attentive children did not. In the cognitive energetic model state factors like arousal, activation and effort are taken into account to explain shortcomings in ADHD patients . According to this model state factors can be moderated by event rate (inter-stimulus-intervals, ISI) and workload in cognitive tasks both under- and over arousal can be produced. Recent research has shown that stimulant medication (methylphenidate) and shortened event rate can produce the same effect in a Go-Nogo task . From our point of view the term arousal is poorly defined in the literature, and could be interpreted in terms of wakefulness or in term of neural arousal. To fully investigate the arousal-noise hypothesis an experiment would have to be designed where physiological arousal is explicitly manipulated and measured. We argue that that proposed framework, including the dopaminergic influence on stochastic resonance, provides a more elaborated view both at the neural and at the behavioral level. To account for the current data an arousal view would have to argue for a selective lower arousal for the inattentive children. Finally, our experience is that the subjects in our experiment are fully aroused, the testing conditions at hand are very stimulating, and subjects are very motivated to perform well.
By highlighting the role of individual differences in the facilitative effects of auditive noise the current study refines our understanding of SR. SR exhibits an inverted U-curve function, where performance peaks at a moderate noise level. However, this is an oversimplification, as there is no absolute sense in which a moderate noise level is optimal. An "optimal" noise level for one individual could be either too high, or too low amount of noise for another individual. These complex interactions between noise and performance may account for some of the contradictory findings in the previous literature. For example, earlier research on noise in normal populations has shown both enhancing and diminishing effects of auditory white noise on cognition in non-clinical groups (90 dB) on simpler, short-term memory tasks like anagrams, whereas speech noise was detrimental . These noise effects also interacted with other variables such as gender and time of the day , which makes these results equivocal. No effect of white noise was found in digit span recall in two experiments, whereas speech noise had a detrimental effect [4, 5]. However, noise improvement was found in a simple addition task in selected groups, elderly and young participants  and among elderly and Alzheimer patients . In Broadbent's early research negative effects of noise have been found using high (excessive) noise levels around 100 dB [62–64]. In later experiments by Broadbent and colleagues, (memory recall of unrelated words) lower noise levels (80-85 dB) were used; results showed no effects of noise on memory when exposed during the encoding phase but deteriorating results if exposed during the recall phase [65, 66]. More recently, episodic memory has been found to be particularly vulnerable to speech noise, whereas traffic noise showed no effect . Results from the present study would indicate that the effects of external noise would have look quite different in many of these studies if participants had been divided into attentive and inattentive or young and elderly, that is high and low gain participants (tentatively high/low dopamine groups). Selective effects of noise can easily get hidden in group-means if some participants improve and others are impaired. Our data may encourage noise researchers to reanalyze their experiments dividing participants by individual differences in attention and performance. Preliminary data from our lab, on an ADHD rat model provide further support for the benefits of adding white noise. The Spontaneous hypertensive rat (SHR) showed improved sensorimotor gating by showing more a pronounced pre-pulse inhibition of the startle reflex when exposed to white noise as compared to control strains, even though control rats also increased their inhibition in noise conditions (Pålsson, Söderlund et al., Noise benefit in pre-pulse inhibition of the acoustic startle reflex, submitted).
Reading disability is a common co-morbidity in ADHD. Consistent with these findings, our data show lower reading skill for the inattentive group. Reading disability is also linked with reduced short-term verbal memory that requires phonetic coding of material, but not necessarily with executive functions or long-term memory [67, 68]. This is also consistent with our data that show a lower performance in the digits forward task that measures short-term memory, but no deficits for digits backward task that is related to working memory capacity. The MBA model accounts for these findings because the digits backward task is a more demanding task leading to higher brain arousal and thus good performance for inattentive children, whereas the less demanding digit forwards task does not sufficiently arouse the brain for the low attention children . Additionally, the positive correlation between reading ability and noise enhancement suggests that white noise may enhance awareness. This is consistent with the idea that dyslexia is caused by phonological deficits , however, a further investigation of phonologic awareness is outside the scoop of the present study.