For the present study we developed a paradigm useful to examine distraction and memory load in a developmental context. Our results indicate that the initial novelty processing of children differs from adults (MMR), that orienting towards the novel is relatively similar between children and adults (P3a) but that reorienting mechanisms are not yet fully developed (RON). On the behavioral level novels surprisingly speeded the children's response whereas they slowed the response in adults.
Both age groups responded faster and more accurate in the low load condition compared to the high load condition. This indicates that the manipulation of the working memory load was successful in both age groups. However, the memory load manipulation revealed no age effects. As expected, children generally responded less accurate than adults.
Novel sounds caused behavioral distraction in adults as reflected by increased reaction times to the visual target when preceded by a task irrelevant novel compared to when preceded by a standard sounds. Similar effects of task-irrelevant novel or deviant sounds on task performance were shown before (e.g. [1, 14]). These effects could not necessarily be expected, as with increasing SOA between sound onset and onset of the target-related information behavioral distraction events caused by deviant sounds get smaller [4, 40]. Most interesting, in contrast to adults, children showed a facilitation effect, that is, responses to visual targets were speeded when the preceding sound was novel. Such a facilitation effect has been reported in adults [41, 42]. San Miguel and colleagues  proposed that (together with other factors) attentional task demands and the temporal position of the novel relative to the encoding or retrieval of the task-related visual information influences whether a novel causes distraction or facilitation. Our study shows that the impact of a novel sound on the performance in a visual primary task is different between school children and adults. Within the theoretical framework of San Miguel and colleagues, we conclude that the alerting potential of novels is larger in children than in adults (at least in the present paradigm). Children do reveal a distraction effect with shorter SOAs between distractor and target information in auditory-visual [18, 43] and auditory-auditory distraction paradigms [19, 21, 24, 25]. Thus, it is possible that it takes some time for the facilitation effect to develop.
A different explanation for facilitation in adults comes from a very recent work by Parmentier et al. . They consider novelty distraction to be modulated by the informational value of the sound. If both task irrelevant sound types, standard and novel, contain the same information about the foreground task (for example about the timing a target will appear), novels cause distraction on behavioral level. But if the novel sounds but not the standard sounds contain information they even result in facilitation. Considering the smaller sensory memory in children  and the large SOA we chose, it may be possible that children could not carry the informational value of the standard sounds to the primary task. The novels, due to their activating nature, may have delivered the timing information more efficiently resulting in a shortage of the reaction time. On the other hand, and in line with Parmentier et al.  adults could have picked up the target information from both stimulus types and thus get distracted from the violation of the auditory pattern.
In the present study, the novelty effect was not modulated by task load (although task load had a clear effect on RTs per se). This adds to the divergent findings with respect to the effect of task load on distraction. It seems that depending on various factors - such as the specific nature of the task or the SOA - an increase in task-load can increase or decrease distraction or it can have no effect.
From a developmental point of view our results are quite astonishing as usually distractors should impair behavior, especially in children. It should be the purpose of future studies to elaborate how this facilitation evolves and which aspects are responsible for it.
The auditory ERPs per se show morphologic differences between children and adults which are consistent with previous studies (e.g. [19, 21, 45, 46]). However, the P3a appears to be relatively similar between the age groups. Thus, although sensory processing, at least its neural basis reflected by the ERPs, is still immature in children, the novel-specific attentional orienting as reflected by P3a is already well developed at the age of 9-10. This is consistent with recent findings about distraction and cognitive control . However, mechanisms underlying RON seem to operate differently in children and adults.
The difference waves of novels and standard sounds in the present paradigm resulted in a positive mismatch response in school children aged 9-10, which previously was only reported in kindergarten children, infants [19, 38, 47], and children aged 7-8 years . Voltage and SCD topographies for this MMR point to a complex generator structure. A prominent central source is accompanied by fronto-lateral sources and parieto-temporal sinks. A combination of temporal areas and deeper central sources appears plausible. Neural sources of the MMR in temporal areas are in line with source modeling done by Maurer and colleagues . Importantly, this MMR was absent in adults and they did also not show an MMN to novels. This absence of MMN was expected on the basis of previous studies showing that MMN is difficult to obtain with omissions of stimulus features [20, 48]. Considering that the environmental sound we used as a standard had a broad frequency spectrum, a novel sound consisted in the omission of parts of the frequency spectrum of the standard. In other words, our stimuli compensate for novelty effects due to different refractory states of novel and standard. We accomplished to control for these effects as was confirmed by our analysis of the N1 time window. These N1 refractoriness effects may in fact be main contributors to the MMN with novels (see  for a review). As such, our results could be seen as further evidence for this interpretation (see also  for similar results). It is still unclear which characteristics are responsible for the elicitation of a positive MMR compared to an MMN. Considering that no other novelty-specific ERP is elicited before the MMR and its latency-similarity to the MMN in adults, it is quite convincing that the MMR reflects the change detection mechanism in the novelty complex . Furthermore, Maurer and colleagues  could show that the MMR is not related to attentional orienting, but is sensitive to the same experimental manipulations as the MMN. However, our results clearly indicate a difference in how novels are processed in children and adults in the time-window of 180-200 ms.
It is still not known which parameters lead to the elicitation of the MMR and, furthermore, its underlying mechanism still has to be investigated intensively. Wetzel et al.  proposed that the long SOA between the sounds is responsible for the frontal positivity. They interpret that the still immature prefrontal cortices lead to a different stimulus processing visible in this positivity. On the other hand, Ĉeponienė et al.  varied the SOA in 7-9 year old children and found neither an attenuation nor amplification of the MMN. But unlike the study of Wetzel et al. , using novels as distractors, Ĉeponienė et al.  used deviant sinusoidal tones. So it appears that the SOA alone is not sufficient to elicit this positivity. It may well be that complexity of the sounds play a more important role.
Supporting this idea, some studies report a quite similar positivity with a latency in the MMN time-range in children ages 9-13 years [17, 18, 37], which is referred to as early P3a (eP3a) circumscribed from the late P3a, also detailed investigated in adults (for a review see ). All mentioned developmental studies used novel sounds and an SOA of at least 1.7 s. The topography of the eP3a is fronto-central and quite focal as is the MMN, only with inversed polarity. Also its generator is thought to be in superior temporal areas, basically the auditory cortex  corresponding well to eP3a generators found in adults [51, 52]. Thus, the positive deflection in our data may be the described eP3a. In the terms of Ĉeponienė and colleagues  it may be that the eP3a governs the attentional shift which is reflected by the late P3a. It is argued that the eP3a is somehow an "attentional-domain 'receiver'", calling for the attentional switch (lP3a) . This is further corroborated by the latency difference of early and late P3a - of around 100 ms which is of the same order needed to shift attention from one spatial focus to another . Also the decreased latency difference in the eP3a-lP3a complex in children compared to adults  coincide with general latency decreases of ERP components found from child- to adulthood . Hence, the similarities of the topographies and scalp current densities of the eP3a in children (see Figure 3 and ) and adults (see ), especially the central source, support the interpretation of both as the same process.
Besides MMR and eP3a there is a third way of how the presented positivity can be interpreted. As its latency is in the P2 range it may be argued that the effect is a modulation of the P2 (see Figure 2). Also the central source in the MMR time window, visible in the SCDs in Figure 3, is supporting this notion. Unfortunately it is even harder to speculate on the underlying function when we regard the MMR as a P2 effect, because the functions underlying the P2 are almost as widely circumscribed as its generators (for a review see ). However, a recent article argued that the P2 mechanism serves to efficiently regulate access to perceptual representations . In this framework our results may indicate that this mechanism in children needs much more effort to access the representation of novels compared to standards, another hint for children being more stimulus driven than adults. In other words, adults may access the perceptual representation of novels and standards similarly, and process the deviating aspect of novels differently thereafter (P3a and RON). In children novels and standards may yield different perceptual representations reflected in the different P2 amplitudes.
Of course we may also think of the MMR as a conglomerate of the described processes including mismatch detection, attention shift government and access to perceptual representations. It is crucial to further explore this positivity in future studies to disentangle its underlying functions and contributing sources.
However, we could show that MMR with a positive polarity at fronto-central leads are not confined to newborns or kindergarten children but can be obtained with children aged 9-10, which show a classical MMN of negative polarity with smaller SOAs and sinusoidal tones instead of novels [20, 21].
Voltage topographies of the P3a show that children's P3a has a more focused, central distribution, whereas the P3a of adults is broader (but also with a central maximum). This is also confirmed by the SCD topographies which show a different pattern of sinks and sources in children, with one central source and two parietal sinks, than in adults, showing only one fronto-central source. Statistics in the P3a interval showed no effect of age on the processing of novels. That leads to the assumption that the mechanism of orienting on task irrelevant stimuli is already well mature at the age of 9-10 years. In that same interval, we showed larger amplitudes in the high load condition in both age groups. This indicates that memory load modulates the brain activity following the task irrelevant sounds. However, this modulation was statistically not dependent on standard or novel sound presentation, i.e. it did not matter whether the sound was violating a pattern or not. So memory load appears to be not mediating the attention switch on task irrelevant novel sounds.
It has been shown, that the P3a is sensitive to working memory load manipulation [12, 27, 28] indicating a different amount of attentional resources oriented on the distracting stimulus under different load conditions. An explanation of our data is to be found in the structure of our task. In this cross-modal design subjects may have been able to focus efficiently on the visual task. Thereby distractor and task stimulus may have been processed in parallel or at least independently. Evidence for this hypothesis comes from behavioral data by Parmentier et al. . These authors propose that attention capture and target processing are independent as long as they are perceptually distinct, which would be the case in our design. Another possibility is delivered by Muller-Gass and Schröger  who show that the amount of attentional resources determine the extent of distraction. In their study stimuli were carrying both, task relevant and distracting information, increasing the amount of attentional resources on the distractor. In our case attentional resources are directed away from the distractor, leaving the same amount of resources in both load conditions for the orienting process reflected by the P3a. A third distinctive feature of our design is the long sound-target SOA of 600 ms. It seems possible that task load did not modulate the attention shift towards the irrelevant sound (reflected by P3a), because the time between the sound and the visual task relevant stimulus was too long. However the novelty complex is clearly affected by load in the time window following the P3a, which will now be discussed.
In the RON time window scalp voltages show the typical frontal pattern in adults, also confirmed by two frontal sinks in the SCDs. In children however, SCDs show additionally to the frontal sinks a prominent central source and two parietal sinks. Additionally, and in contrast to the P3a, the RON was delayed in children compared to adults. This extends the finding reported by Horváth and colleagues  for 6 year old children to a different age group. Furthermore it strengthens the idea of a functional independence of P3a and RON. Future studies should investigate the development of this delay from early to late school age. Contradictory to the MMR and P3a, RON is clearly affected by the memory load manipulation in adults whereas it is not in children. This results support recent findings  showing that children 7-8 years old are more stimulus-driven than adults. While in the present study in adults the reorienting process is diminished under high memory load, it elicits the same response in children independently of the current load. The findings in adults also support results by Berti and Schröger , who found a decreasing RON with increasing memory load. Following their argumentation we can assume that adults were able to highly focus on the demanding task thereby decreasing the resources admitted to involuntary attention switches, here re-orienting of attention to the task. It was shown before that these executive functions within working memory, which are necessary for the balance between voluntary and involuntary attention resources, are less developed in children especially when it comes to highly salient distractors as used in the present paradigm. As the RON in the children's group is similar in both load conditions we assume that the underlying process of reallocation of attention to the task is carried out to the same extent.
Our divergent results in the P3a and RON time windows are confirmed by recent studies which questioned the strength of the link between P3a and RON in adults [56–58]. Furthermore, comparing children aged 7-8 and adults Wetzel and colleagues  also find dissociations between P3a and RON. Together with our results this indicates that the underlying mechanisms of P3a and RON are not only partly independent but also developing at different ages. It appears that the involuntary attention shift is already well developed around 9 - 10 years while the reorientation mechanism is not.