Riding the elephant in the room: Towards a revival of the optimal level of stimulation model

Phenomena such as engagement, attention and curiosity rely heavily on the “ optimal-level of stimulation (or arousal) ” model, which suggests they are driven by stimuli being neither too simple nor too complex. Two points often overlooked in psychology are that each stimulus is simultaneously processed with its context, and that a stimulus complexity is relative to an individual ’ s cognitive resources to process it. According to the “ optimal-level of stimulation ” model, while familiar contexts may decrease the overall stimulation and favour exploration of novelty, a novel context may increase the overall stimulation and favour preference for familiarity. In order to stay closer to their optimum when stimulation is getting too high or too low, individuals can explore other stimuli, adopt a different processing style or be creative. The need and the ability to adopt such strategies will depend upon the cognitive resources available, which can be affected by contextual stimulation and by other factors such as age, mood or arousability. Drawing on empirical research in cognitive and developmental psychology, we provide here an updated “ optimal-level of stimulation ” model, which is holistic and coherent with previous literature. Once taken into account the role of contextual stimulation as well as the diverse factors influencing internal cognitive resources, such model fits with and enriches other existing theories related to exploratory behaviors. By doing so, it provides a useful framework to investigate proximate explanations underlying learning and cognitive development, and to develop future interventions related, for example, to


Introduction
From birth, humans actively engage in the construction of their own knowledge by exploring their environment and experiencing the resulting habituations and conditionings.For this reason, exploratory behaviours and the factors determining them are implicitly at the heart of cognitive development research.These factors, be they genetic or environmental, can shape learning trajectories leading the developing child towards typical or atypical routes (Karmiloff-Smith, 2018;Westermann et al., 2007).Nevertheless, despite the extensive existing literature on the topic, we still do not fully understand when and why key factors such as familiarity or novelty preferences occur (Mather, 2013).
Historically, two influential models have been proposed: the mere exposure effect to explain familiarity preference, and the habituation model to explain novelty attraction.However, there are core inconsistencies between these two models.According to the mere exposure effect, a repeated exposure to a stimulus increases its positive emotional valence, and then leads to a familiarity preference.However, while the mere exposure effect accounts for familiarity preference, it does not explain (1) why boredom can limit the pleasantness provoked by familiar stimuli (Baillargeon et al., 1985;Berlyne, 1960;Bornstein, 1989;Stang, 1975), and (2) why developmental differences exist in preference for familiarity and novelty (Bornstein, 1989).The phenomena of boredom and of children's interest in novelty are more consistent with models of habituation which, in direct contrast to the mere exposure effect, argue for a decrease of interest in familiar stimuli with repeated exposure, and a preference for novelty (Fantz, 1964;S. Friedman, 1972;Quinn & Eimas, 1986;Rieth & Sireteanu, 1994;Saayman et al., 1964;Slater et al., 1982Slater et al., , 1984;;Sokolov, 1963;Thompson & Spencer, 1966).
The optimal-level of stimulation (or arousal) model, which hypothesizes that individuals actually seek and prefer an optimal (usually moderate) level of stimulation has generally been accepted as a plausible explanation for the seemingly contradictory results stemming from the mere exposure and the habituation literatures (Berlyne, 1960(Berlyne, , 1970;;Colombo & Mitchell, 2009;Kaplan et al., 1990;Mather, 2013;Montoya et al., 2017).Other very similar models such as the two-factor model of mere exposure, and the dualprocess theory make the same predictions (Bornstein, 1989).The two-factor model conceptualizes the mere exposure effect in terms of the combined effects of habituation, which makes new stimuli easier to process and less threatening, and boredom, which results in the decline of positive affect.The dual-process theory infers two interacting processes triggered by every stimulus: habituation which increases familiarisation towards a specific stimulus and sensitization which determines the global level of arousal (Groves and Thompson, 1970).While one increases, the second may decrease resulting in varied possible outcomes.
Finally, others have proposed models based on information processing.From that perspective, information is subjected to the 'Exploitation-Exploration Dilemma'.Such model suggests that a current source of information is processed (exploited) until the effort needed to extract additional information outweighs the effort needed to seek information (explore) elsewhere, at which point a shift in the direction of attention occurs (Hills et al., 2015;Laureiro-Martínez et al., 2015;Piccardi et al., 2020).It is worth noting that by considering information as a source of stimulation, the "optimal-level of stimulation" model offers a promising proximate explanation when facing such dilemma.In a nutshell, individuals in such a situation could be comparing the stimulation potential of the varied options.For example, if exploitation gets too boring, exploring alternatives may become more attractive.
All these models appear to be consistent tools for understanding exploration as well as subsequent learning and developmental trajectories.This review article will focus on the optimal-level of stimulation model.It aims to propose a synthesis of related concepts and factors usually treated separately in the literature.Such systemic approach can increase the explanatory power of the optimal-level of stimulation model and allows us to generate a series of testable predictions for consideration in future research.
After a brief consideration of semantic issues which hide the central role of stimulation, we will introduce the mechanisms related to the mere exposure and habituation model which could explain approach towards familiarity or novelty, and withdrawal from novelty or familiarity.We will then present the optimal-level of stimulation model and examine how considering contextual variables as well as diverse key factors influencing internal cognitive resources can enrich it.Finally, we will discuss how this updated model can contribute to existing theories related to attention, exploratory behaviours and cognitive development, and its potentials for real world applications.

Preference
In previous studies, the term "preference" as in "preferential looking experiments" is often used when babies invest more time in exploring certain stimuli over others (Cooper & Aslin, 1990;Fernald, 1985;Hamlin et al., 2013;Hamlin & Wynn, 2011;Moon & Fifer, 1990;Slater et al., 1998;Wattam-Bell, 1996).This is unfortunate since such a term carries a positive connotation that may be misleading.In fact, there is no evidence that babies in those studies really had a preferencea positive attitude or a greater likingtowards the stimuli they were exploring.In fact, exploration may not necessarily be linked to positive attitudes, and may even be linked to negative ones (Harrison, 1968), as it is the case, for example, when we pay attention to any disgusting or potentially harmful stimulus.Moreover, when the reasons for the direction of infants' attention is of theoretical interest, the misleading term "preference" can have profound consequences for the hypothesis under investigation (Bogartz et al., 2000;Cashon & Cohen, 2000;Houston-Price & Nakai, 2004;Schilling, 2000;Sirois & Jackson, 2007).For these reasons, we will instead henceforth use more neutral terms such as approach and withdrawal which are commonly used in the reward literature to reflect the seeking component, in contrast to "liking", the hedonic component of the stimulus (White, 2011).

Complexity, novelty, intensity and surprise
Terms such as complexity, novelty, intensity or surprise are also problematic since they confound the interaction between the properties of a stimulus and the expectations and cognitive resources of the perceiver (Giacalone et al., 2014;Palczak et al., 2019;Sherman et al., 2015).For instance, a rattle can appear extraordinarily complex and novel for a baby, while it would be trivial and familiar for a toddler.
Similarly, novelty, intensity or surprise may depends on the subject's past experience with the stimulus, his/her expectation, and his/ her level of processing.So, considering the role of subjective perception, complexity, novelty, intensity or surprise cannot be considered as fixed properties which make these concepts slippery and extremely hard to measure (Berlyne, 1971;Nadal et al., 2010;Palczak et al., 2019).For this reason, they need to be grounded on some ultimate criterion.
Complexity and novelty, along with intensity and surprise, do have the common point of being the main determinants of cortical stimulation and arousal (Berlyne, 1960;Gray & Eysenck, 2017).Stimulation or arousal, being susceptible to be measured objectively through neuro-physiological changes (e.g., pupil dilation, heart rate or electro-dermal activity), appear as much more functional constructs.For this reason, we will instead henceforth use the broader term "stimulation potential" when talking about the level of stimulation provoked by the stimulus or by its context, following the terms used by arousal theorists and "sensation seeking" theories (Berlyne, 1960;Zuckerman, 1994).
Other evidence come from intermodal studies with infants looking longer at a visual stimulus that matches an auditory stimulus (more familiar) over one that mismatches it (Golinkoff et al., 1987;Spelke, 1976).In adults as well, congruent sounds increase the pleasantness of olfactory stimuli (Seo et al., 2014;Seo & Hummel, 2011).

Approach towards novelty and withdrawal from familiarity
In sharp contrast with the literature described in the previous section, we can also be attracted by novelty and we can invest large amounts of time and effort in exploratory behaviours (Rusher et al., 1995;Weisler & McCall, 1976).Indeed, exploration is considered the hallmark of our curiosity and the source of our learning.Such need for novelty is even described as an essential factor in personality theories with traits such as "sensation seeking" (Zuckerman, 1994) and "openness" (Eysenck, 1985).
Interestingly, participants with high boredom susceptibility have been found to habituate more quickly to repeated presentations (Jiang et al., 2009).In fact, boredom experience has been linked to habituation of cortical arousal in response to repetitive stimulation (Hamilton, 1981;O'Hanlon, 1981;Zuckerman, 1994).In this case, attention towards unfamiliar or unexpected stimuli would reflect their higher stimulation potential (see Fig. 1).

Optimal-level of stimulation and exploratory behaviours
By manipulating the number and the length of familiarization trials, psychologists noticed that a brief exposure can make a stimulus attractive while longer familiarization makes it repelling (Hunter & Ames, 1988;Wagner & Sakovits, 1986).Other studies with infants confirmed this finding with incomplete habituation (i.e., several exposures, but too few to reach habituation) leading to longer looking times and exploration for semi-novel objects (Colombo & Bundy, 1983;Hunter et al., 1982Hunter et al., , 1983;;Kidd et al., 2012;Mather, 2013;Roder et al., 2000;Rose et al., 1982).This is the reason why researchers are advised to use infant-controlled habituation procedure to be sure their participants fully habituate (Slater et al., 1984).In the same vein, labelled the "goldilocks effect", infants allocate more attention to intermediate levels of complexity (Kidd et al., 2012(Kidd et al., , 2014)).A more recent study confirmed that adult participants also show more positive evaluations for semi-novel stimuli compared with very familiar or very novel ones (Gustafsson et al., 2021).
Such attraction towards semi-novelty (or semi-familiarity) has been explained mainly through the "discrepancy hypothesis" (e.g., Dember & Earl, 1957;McCall et al., 1977;McCall & McGhee, 1977).According to it, the attentional shift from familiar towards novel stimuli is thought to occur when the encoding of the familiar stimulus is complete, that is, when there is no discrepancy between the stimulus input and its internal representation (Kagan & et al, 1971;McCall & Kagan, 1967;McCall & McGhee, 1977;Pascalis & De Haan, 2003).The preference for processing semi-novelty would be caused by a residual activity in memory acting as a "proto-schema" until stimuli are stored in the long-term memory (Bogartz et al., 1997).Attentional shift towards an optimal-level of stimulation (or arousal) has generally been accepted as a plausible explanation for the seemingly contradictory outcomes of mere exposure and habituation studies (Houston-Price & Nakai, 2004;Hunter & Ames, 1988;Mather, 2013; see Fig. 2).
The level of deviation between the actual and the optimal stimulation would determine the agent's intrinsic motivation to explore, and the type of exploratory behaviours displayed (Raju, 1980;Steenkamp & Baumgartner, 1992).The greater the deviation is between an individual's actual stimulation level and his/her optimal level, the greater the extent to which he/she will engage in exploratory behaviours, either towards familiarity or novelty.For instance, boredom can promote mind-wandering and creativity, which may be considered as strategies to cope when the environment only provides very few affordances (Forster & Lavie, 2009;Gustafsson, 2022;Mann & Cadman, 2014).In contrast, environmental overstimulation may drive our attention towards more familiar stimuli.For example, in infancy, very familiar and portable items such as the security blanket can help deal with overstimulation and allow for an optimal pace of exploration (Passman & Weisberg, 1975).
The need for an optimal-level of stimulation has been described as a basic psychological need (González-Cutre et al., 2016), and infant studies strongly suggest that such mechanism is well in place and functional from birth, determining exploration and learning opportunities (Hunter & Ames, 1988;Mather, 2013;McCall & McGhee, 1977).Interestingly, in any given environment, individuals can differ greatly in their propensity for seeking stimulation (Raju, 1980;Steenkamp & Burgess, 2002).The optimal-level of stimulation model can also help to explain such inter-individual variability.In fact, if the fluency of information processing is so deterministic in predicting responses towards a given stimulus, we can hypothesize that any factors affecting sensory stimulation and the cognitive resources available to process that information such as context, age, mood and arousability should be considered to predict exploration outcomes.

Contextual stimulation potential
A point that is often overlooked in psychology studies is that each stimulus is always simultaneously processed with its context (de Zilva et al., 2016;Dissegna et al., 2021;Orne, 1962;Rosenzweig, 1933).Assuming that processing stimuli is costly, the stimulation potential of the context in which the stimulus is embedded (i.e., its novelty, complexity, intensity or surprise features) may also play a role in exploratory behaviours (see Fig. 3).According to the optimal-level of stimulation model, we can then hypothesize that, while familiar contexts may decrease overall stimulation they will favour exploration of novelty.In contrast, while novel contexts may increase overall stimulation they will favour exploration of familiarity.
This point would also explain why a strong apparent "preference" for familiarity can easily happen in experimental conditions, as demonstrated in "mere exposure effect" studies.Since the experimental context is highly stimulating for the participant, the less stimulating familiar stimulus could offer the opportunity to stay optimally stimulated, explaining the positive attitude towards such stimulus.In a similar vein, the striking inclination of newborns towards familiarity could be explained by the extremely novel and over-stimulating environment they found themselves in just after birth.Focusing on familiar stimuli (e.g., mother's face, odour or voice) could allow them to stay closer to their optimal level of stimulation (Mastropieri & Turkewitz, 1999;Pascalis et al., 1995;Schaal et al., 1998Schaal et al., , 2000)).
Several lines of evidence support this idea.First, a recent meta-analysis on "Head Turn Preference Procedure" (HPP) studies pointed out that infants with more experience with such experimental setting were less likely to show an approach towards familiarity than those who had less experience with it (Santolin et al., 2021).Another evidence comes from a study by Lewkowitz and Turkewitz (1981) who showed that, while usually preferring a light with moderate intensity (in line with the optimal-level of stimulation), infants tended to prefer the light with the lowest intensity after being stimulated by a noise.This suggests again that infants can modulate their intake of information according to the level of contextual stimulation in order to stay closer to their optimum.
If exploratory behaviours are modulated by the interaction between an individual's unique need for stimulation, and the stimulation potential of his/her environment (stimuli and context), it is important to also consider that person's characteristics.When investigating intrinsic motivations to engage in exploratory behaviours, one determining characteristic is age (Berlyne, 1966;Henderson, 1984;Henderson & Moore, 1980).

Age
Arousal theorists hypothesized very early that, for a given age, an intermediate level of stimulus complexity should be preferred, and that this level should increase as the infant becomes older (Dember & Earl, 1957).Accordingly, preterm infants show diminished habituation performance (Kavšek & Bornstein, 2010), and the older the infants, the more quickly they habituate and develop an approach towards novelty (Hunt, 1970).For instance, while newborns tend to explore familiarity more (for odours: Cernoch & Porter, 1985;Schaal et al., 1998Schaal et al., , 2000;;  food: Dovey et al., 2008), older infants and young children explore and inspect novelty more (Dovey et al., 2008;Fantz, 1964;Hunter et al., 1982Hunter et al., , 1983;;Nachman et al., 1986).Such progression would reflect the increased information processing speed, permitted by the broadening of mental categories and cognitive resources acquired through experience and brain maturation (Brennan et al., 1966;DeLoache et al., 1978;Greenberg & O'Donnell, 1972; see Fig. 4).This developmental pattern can be attenuated by increasing the stimulus complexity, the stimulus salience or the task difficulty (Hunter & Ames, 1988;Wagner & Sakovits, 1986).
Interestingly, there have been many attempts in the literature to measure objective complexity alongside subjective complexity and relate them to aesthetic appreciation (See Van Geert & Wagemans, 2020 for a review, or Van Geert &Wagemans, 2021 andGüçlütürk et al., 2016 for recent attempts to investigate the question experimentally).While some participants show a decrease in liking with increased complexity, others show an increase in liking for stimuli with increased complexity.To explain such opposing results, the Pleasure-Interest Model of Aaesthetic Liking (PIA Model) has been proposed (Graf & Landwehr, 2015).It states that an aesthetic object may be processed in two stages.First an automatic processing takes place, and then if the viewer is motivated enough to process the stimuli further, a controlled processing follows.Similar to the Optimal-level of stimulation model, the PIA Model predicts that mere automatic processing of stimuli would result in an increase in liking, as the stimulus complexity increases, followed by a decrease of liking if the complexity levels are high enough to cause confusion.In other words, this model highlights that alongside a participant's prior experience, perceived complexity may also depend on the participant's motivation, capacity and opportunity to go beyond automatic processing in order to process consciously the stimulus and extract all the information it may contain.All these variables can vary dramatically with age.
Finally, it is intriguing to note that older children and adults explore globally less than young children (Gopnik et al., 2017;Sumner et al., 2019).This may appear counter-intuitive as we may assume they have more cognitive resources, and so a higher optimal-level of stimulation, requiring more exploration to satisfy it.However, although they may have more cognitive resources, it is not clear if these resources are available for exploration.In fact, it is believed that younger children explore more because the costs of everyday survival are borne by carers (Carruthers, 2002;Gopnik et al., 2017;Piantadosi & Kidd, 2016;Tomasello, 2019).So adults have to face trade-offs between the motivational system driving curiosity and other motivational systems driving for example food seeking, mate seeking, and kin care (Ko et al., 2020).In line with the PIA model, it is also possible that they can extract more information from existing stimuli decreasing the need to look for more.
According to the "Regulatory Focus Theory" (Higgins, 1997), when exploring our environment, two separate motivational orientations can be distinguished, namely a promotion focus -in which people focus on growth-and a prevention focus -in which people focus on maintaining security.Promotion state has been linked to a broadening of mental categories by adopting a more global processing style, while prevention state has been linked to their narrowing (Fredrickson, 2001;Friedman & Förster, 2000, 2001).Accordingly, novelty is more attractive in a promotion state (Gillebaart et al., 2012).In fact, broadening mental categories increases the likelihood that new information integrates pre-existing representations, is processed fluently and appears familiar (Förster, Marguc, & Gillebaart, 2010;Liberman, Trope, & Stephan, 2007;Rosch, 1975; see Fig. 5).In contrast, familiarity is more appealing in a prevention state (Gillebaart et al., 2012); and indeed anxious or depressed individuals, who frequently display a hyperactive prevention focus (Clark & Wells, 1995;Klenk et al., 2011;Rapee & Heimberg, 1997), are often characterized by diminished exploration of novelty (Carton et al., 1992;Farmer et al., 2001;Kashdan, 2007).
A promotion or a prevention focus motivation can be primed either by the emotional valence of a stimulus or by the emotional valence of the context in which it is presented (Freitas et al., 2005;Gillebaart et al., 2012;Marguc et al., 2011).For example, the pleasantness of ambient scent and shopping environments can influence exploratory behaviours (Mattila & Wirtz, 2001;Orth & Bourrain, 2005;Wakefield & Baker, 1998;Wirtz & Mattila, 2008).Interestingly, positive contextual information can also alter appraisals, by changing the person mindset.For example, in related studies, participants preferred people original art compared to (identical) forgeries (Newman & Bloom, 2012), images labeled as gallery exhibits instead of computer-generated (Kirk et al., 2009), images labeled as art instead of non-art (Wagner et al., 2014), branded product instead of not-branded (McClure et al., 2004), or even dot patterns presented as stars instead of peas (Hansen & Topolinski, 2011).
Lastly, it is also important to note that the emotional valence of a stimulus and its context interacts with mood and cognitive resources in a dynamic way.In line with the dual-process theory mentioned before, habituation can create a transient increase of interest in a repeated stimulus, but the underlying increase in arousal is not stimulus-specific (in contrast to the stimulus specificity of habituation).Under this interpretation, sensitization by increasing arousal can increase interest in either another more familiar stimulus if the arousal was depleting internal resources, (e.g. in case of a stressful stimulus), or in a more novel one if the arousal was rewarding/energizing.For instance, seeing something disgusting may attract attention at first because such stimulus is particularly arousing in a familiar environment, making us closer to our optimal level of stimulation.Afterwards, the negative emotions associated with it may decrease the cognitive resources available to process it and lower our optimal-level of stimulation.This will then drive exploratory behaviours towards less arousing stimuli which would be closer to the new optimum.
Another example could be found within the Hedonic adaptation prevention model (Sheldon & Lyubomirsky, 2012).According to this model, positive emotions triggered by life changes will fade with time unless we find alternative ways of benefiting from these changes or unless we focus our attention on them in order to favour a continued appreciation.In line with our updated optimal level of stimulation model, the rewarding effect of a positive change such as a new Christmas toy will skew the child optimum towards exploration.With time such positive arousal will decrease as well as the rewarding effect of the new toy (e.g. by eliciting less praise from peers, by providing fewer affordances once all the functionalities are discovered, etc…).In parallel, as the toy becomes more familiar, its stimulation potential will stray further away from the child's optimum.At this point, finding alternative uses, or Fig. 6.Updated Optimal-level of stimulation model, including contextual stimulation potential, cognitive resources available and arousalility.consciously increasing the appreciation for the toy could help maintain the positive emotions initially triggered.In line with the PIA Model previously mentioned, conscious processing may also help extract more information and increase the subjective complexity of the stimulus.

Arousability
According to Eysenck's arousal theory, the ascending branch of the reticular activating system allows more sensory stimuli to pass onto the cortex in introverts than in extraverts (See Fig. 6).Consequently, introverts are more easily aroused, explaining why they tend to limit stimulation.Conversely, extraverts would be often under-stimulated and may seek out further stimulation (Corr, 2004;H. Eysenck, 1967).Accordingly, extraversion has been found to correlate positively with the personality traits "sensation seeking" and "openness" (Aluja et al., 2003;Haapasalo, 1990;Zuckerman, 1994;Zuckerman et al., 1978).In fact, high sensation-seekers require more stimulation to reach their optimum and are therefore more open to new experiences (Pliner & Melo, 1997).

Reinforcement sensitivity theory
This theory postulates three systems: (1) the fight-flight-freeze system (FFFS), responsible for avoidance and escape behaviours by responding to cues of punishment/non-reward; (2) the behavioural activation system (BAS), which regulates approach behaviours by being responsive to signals of reward; and (3) the behavioural inhibition system (BIS), responsible for resolving conflict between FFFSavoidance and BAS-approach, which can generate anxiety (Corr, 2004;McNaughton & Corr, 2004).
One contribution from the optimal-level of stimulation model is to highlight how the stimulation potential of a stimulus (and its context) can influence the BAS and the BIS mechanisms.In fact, according to the optimal-level of stimulation model, in order to trigger the intrinsic motivation to explore, a stimulus must have a positive emotional valence which can also result from its stimulation potential (Cahill-Solis & Witryol, 1994;Subbotsky, 2010) in addition to previous reinforcements.Very early on, both Schneirla (1959) and Berlyne (1971) suggested that stimuli with optimal stimulation potential may engage reward mechanisms in the brain.For instance, a very simple stimulus could have a positive valence in the eyes of a bored person in an empty environment.This is in line with the persistent evidence that the dopamine system can react specifically to unusual stimuli which do not involve any primary reward (Bunzeck & Düzel, 2006;Costa et al., 2014;Düzel et al., 2010;Kakade & Dayan, 2002;Reed et al., 1996).Interestingly, this can also come from internal physical or mental processes.For instance, a bored person could engage in repetitive fidgeting movements, active thinking or daydreaming for the sake of the pleasurable stimulation such activity provokes.
In contrast, stimuli or situations having sub-or over-optimal stimulation potential could be perceived as boring and be treated as negative reinforcer.In fact, 'boredom' is arguably more than just a decrease of interest with repeated exposure.It is an aversive state that occurs when we are not able to successfully engage our attention with our environment (Eastwood et al., 2012).So, any difficulties encountered in regulating the inflow of information to keep the organism optimally stimulated would result in some form of boredom.This can occur when all the surrounding stimuli remain too complex (e.g., a math lesson) or too simple (e.g., waiting for the bus with no book or phone) in relation to our optimal-level of stimulation.
Interestingly, the BAS scale used to measure "approach motivation" correlates positively with boredom susceptibility, and the BIS scale used to measure "avoidance motivation" is positively correlated with boredom proneness, in boring situations (Mercer-Lynn et al., 2013, 2014).People scoring high in the BAS scale, may be more susceptible to boredom because they are often under-stimulated regardless the type of environment (Eysenck, 1967;Mercer-Lynn et al., 2014, 2013) or because they expect and seek out bigger rewards leading to a lower subjectively rewarding experience compared to a low BAS person in the same situation (Pickering & Smillie, 2008).People scoring high in the BIS scale would be more susceptible to boredom in boring environments because sub-optimal stimulation in such environment may act as a punishment.
Counterintuitively, it is also worth noting that humans can sometimes actively seek very high arousal states.Berlyne (1960) hypothesized that such "arousal jags" are pleasurable only because of the anticipated drop in arousal activity, like when riding roller coasters.If this is the right explanation, such a jag would not be pleasurable the first time and would require an initial conditioning.This might explain why they are not observed in newborns.The development of tickling play and how the child can proactively start it could provide an interesting situation to investigate this question (Provine, 2001).
The "Learning Progress" hypothesis According to the "learning progress" hypothesis, "…the brain, seen as a predictive machine constantly trying to anticipate what will happen next, is intrinsically motivated to pursue activities in which predictions are improving, i.e., where uncertainty is decreasing and learning is actually happening."(Gottlieb et al., 2013;Oudeyer et al., 2016).In fact, in a state of uncertainty or in a state of need, we may avidly seek out comprehensible information to achieve a goal.The high degree of motivation associated with such behaviour suggests that, indeed, learning may be rewarding per se.However, this theory does not explain why improving predictions, i.e., having our expectations fulfilled, is intrinsically rewarding.
Current research suggests that expectations or predictions happen when the brain bias the patterns of neuronal activity in the way that is most likely to be appropriate when encountering the stimulus (Meirhaeghe et al., 2021).Such biased brain activity would likely make new information processing more fluent, resulting in an intrinsically rewarding effect (in line with the hedonic fluency model mentioned before).In fact, a recent computational model suggests that the better the stimulus can be predicted based on the preceding one(s), the easier it is to process, and so the greater the fluency (Brielmann & Dayan, 2021).These authors propose to re-phrase fluency theories as stating that "sensory experiences associated with lower prediction errors are more pleasurable".However, at this point, one could argue such approach would face the same limitations as the mere exposure model ones (i.e., why can very familiar or predictable stimuli be perceived as boring?).To address this issue, the authors' model predicts that people find sensory experiences valuable not only when they are relatively fluent (and therefore predictable) but when they also change the neural system in such a way that it predicts future sensory input more accurately.In other words, sensory experiences associated with lower prediction errors are more pleasurable when they are not too trivial, that is when they fall in the optimal-level of stimulation range.
If fulfilling expectations can be rewarding, it could also be that certain violations of expectations may be overstimulating and so perceived as unpleasant.This may be especially true in the early years when expectations are regularly violated.Interestingly, the sight of an object that violates expectations has been shown to enhance and promote information-seeking behaviours in infants (Stahl & Feigenson, 2015).We could then consider exploratory behaviours as strategies developed, through experience, to get a better sense of control on our level of stimulation.Later on, such learning, by being associated with exploration, could reinforce positively "exploratory states".

Load theory of selective attention and cognitive control
According to Lavie and colleagues (2004), while a high perceptual load reduces distractors' interference (e.g., zoning out in front of television), a higher cognitive load increases distractors' interference (e.g., struggle to focus in class).Following this observation, the authors hypothesized the existence of a "perceptual selection" mechanism and a "cognitive control" mechanism that would be necessary to maintain our attention focus on a specific source of stimulation.Such perceptual and cognitive filters would allow us to focus our attention.If the cognitive load increases because of the level of stimulation or because of a low physiological state such as fatigue, we would not be able to maintain the filters anymore, therefore resulting in higher distractibility.
The optimal-level of stimulation model, while acknowledging the important role of cognitive load in determining attentional orientation and exploratory behaviours, offers a more parsimonious explanation, since it does not require any hypothetical perceptual or cognitive filters.It would nuance previous theories by highlighting that it is an optimal rather than a high level of perceptual load that would lead to the reduction of distractor interferences.In other words, distractibility would be an artefact due to not being able to engage our attention with any stimulation closer enough to our optimum, rather than reflecting an incapacity to maintain hypothetical perceptual selection and cognitive control mechanisms.For instance, the optimal-level of stimulation model would predict higher distractibility if the stimulation is too trivial or complex.(e.g., in this case, zoning out in front of television may not be possible if the program is perceived as too simple or too complex).Finally, physiological state such as fatigue may decrease the cognitive resources available, resulting in increased receptivity to less stimulating stimuli which could be interpreted as distractibility.
We hypothesize that previous studies from Lavie did not show any negative effect of a too high perceptual load simply because they did not provide a sufficiently high level of stimulation.In fact, much of these studies involved artificial situations presenting letters on a screen as a stimulus, often in conjunction with distractor letters (Lavie et al., 2004;Lavie & De Fockert, 2003).To strengthen our point, literature related to sensory overload highlights that high distractibility, decreased attention and poor concentration, are the main observable signs of sensory overload (Scheydt, Needham, Nielsen, & Behrens, 2016;Scheydt et al., 2017).

Self-determination theory
The self-determination theory proposes that the needs for autonomy (i.e., need to experience actions as personally chosen), competence (i.e., the need to gain self-referenced success on tasks), and relatedness (i.e., the need to feel understood and valued by others) are basic psychological needs determining intrinsic motivation and individual well-being (Ryan & Deci, 2017;Sheldon et al., 2001).
Another candidate need that has recently been proposed is the need for novelty (González-Cutre et al., 2016, 2020).Although such construct seems to fit with most of the inclusion criteria proposed for defining a basic psychological need, one of the criticisms is that an excess of novelty is not necessarily positive and therefore it cannot be considered as a basic psychological need.Reframing the basic psychological need for novelty as a need for an optimal-level of stimulation would address this issue (i.e., the satisfaction of this basic psychological need would be obtained if people reach an optimal level of novelty and not a high level of novelty).Moreover, by highlighting the interaction between an individual's unique need for stimulation, and the stimulation potential of his/her environment (stimuli and context), the updated model presented here would help address confounding contextual, cultural and individual factors when investigating the universality of such basic psychological need.

Neuroconstructivism and modularity of mind
Previous research has theorized that the innate attraction towards inverted high-contrast blobs triangle could bootstrap face processing and support the development of social cognition (Farroni et al., 2005;Sirois & Karmiloff-Smith, 2009;Sirois et al., 2008).In the same way, newborns look longer at upright biological motion displays which could bootstrap processing of social cues (Bardi et al., P. Ibáñez de Aldecoa et al. 2011;Simion et al., 2002Simion et al., , 2008)).Previous works hypothesized these results reflect the presence of innate modules or inborn hardwired neural systems as part of an evolutionarily ancient system selected to pay attention to such pattern (Bardi et al., 2011;Baron-Cohen, 1997;Baron-Cohen, 1997;Batki, Baron-Cohen, Wheelwright, Connellan, & Ahluwalia, 2000;Simion et al., 2008).In addition to these wirings, we hypothesize that evolution could have also selected for a specific amount of cognitive resources available from birth that would match the complexity of the most relevant patterns found in a newborn's environment, making them closer to the infant's optimal level of stimulation.In fact, the optimal-level of stimulation model suggests that infants' orientation towards such patterns may actually just reflect the optimal-stimulation potential of these stimuli for most babies.The question of "why inverted blobs traingles and upright biomotion displays are more attractive than upright blobs triangles and inverted biomotion displays?"could then be rephrased as "how these patterns stimulate newborns optimally"?
An interesting avenue for future research is to further test such evolutionary hypothesis with other sensory modalities.For instance, does the arousal potential of breastmilk odour match the newborn optimal-level of olfactory stimulation?Does the arousal potential of motherese speech match a baby optimal-level of auditory stimulation?Such findings could have dramatic implications.If some babies are very arousable or oversensitive, they may not pay attention to certain stimuli, even if they are crucial to bootstrap their social development, such as the inverted blobs triangles.They may rather turn their attention towards less stimulating external or internal stimuli, which would be closer to their own optimum but could bootstrap their cognitive development in a different direction.It is worth noting that the idea that at least some of the deviant behaviours displayed by hyperactive or autistic children are attempts to correct "chronic imbalances in arousal" are not entirely new in the scientific literature (Zentall & Zentall, 1983).For instance, autistic disorder has been characterized by arousal regulation problems, sensory modulation difficulties, and atypical attention (Gillott et al., 2001;Orekhova & Stroganova, 2014), as well as withdrawal from the external world, preference for more local processing (Happé & Frith, 2006;Koldewyn et al., 2013) and resistance to change (APA, 2013).

Neophobia on the table. Picky eating and food fussiness
An aspect of withdrawal from novelty, that can potentially have a negative impact on children's development, is food neophobia (Cano et al., 2016;Dubois et al., 2007).Food neophobia is usually interpreted as an adaptive strategy to limit ingestion of dangerous substances that might be mistaken for food (Glendinning, 1994;Jaeger et al., 2017).While it can be adaptive by protecting from potential harm, neophobia can also be the source of eating disorders and nutritional deficiencies later in life (Cashdan, 1994;Dovey et al., 2008;Guzek et al., 2017;Lafraire et al., 2016;Russell & Worsley, 2008).
In line with the optimal-level of stimulation model, exposure during infancy increases the appeal of a novel food (Appleton et al., 2016;Birch, 1998;Birch et al., 1987;Carruth et al., 2004;Gerrish & Mennella, 2001;Maier et al., 2007;Maier-Nöth et al., 2016;Mennella et al., 2008) but excessive familiarity may lead to "monotony" (Rozin & Vollmecke, 1986;Wadhera & Capaldi-Phillips, 2014).Other studies comparing food complexity and hedonic responses found either a negative relationship, or were inconclusive (Palczak et al., 2019).Such inconsistent findings have been explained by the failure to include the full range of subjective complexity.Another aspect highlighted by the updated optimal-level of stimulation model proposed in this article is the role of context.For instance, as noted previously, the experimental context is something unusual that can increase artificially the participant's stimulation level (Rosenzweig, 1933;Rosnow, 2002).This could explain why negative relationships between food complexity and hedonic response were found more frequently.
Future experiments could control for such contextual priming and stimulation potential, and test explicitly the model proposed in this article.For instance, what would happen if unfamiliar foods were presented in a familiar vs an unfamiliar environment?Very few studies controlled for such contextual variables.Some works tested the role of positive contexts and showed that visually appealing presentations could favour food consumption (Jansen et al., 2010).Similarly, foods paired with positive information are more likely to be accepted (Appleton et al., 2016;Martins, Pelchat, & Pliner, 1997;Pelchat & Pliner, 1995); while authoritarian and coercive practices may have the opposite effect (Rigal et al., 2012).Other studies showed children accepting new foods more easily after observing significant others (Addessi et al., 2005;Bevelander et al., 2013;Birch, 1980;Harper & Sanders, 1975;Hendy, 2002;Hendy & Raudenbush, 2000;Highberger & Carothers, 1977;Laureati et al., 2014;Salvy et al., 2008;Shutts et al., 2009Shutts et al., , 2013)).
Future studies could manipulate the stimulation potential of the physical context (e.g., presenting food in a richly noisy decorated room vs on a bare table in a silent room), and the emotional context (e.g., having encouraging parents vs quiet ones) and see how the interaction between these two variables could affect food exploration in paired choice tests between novel vs familiar foods.This could inform interventions aimed at increasing the consumption of novel foods in neophobic children, and the consumption of familiar foods in picky ones.

Anxiety and distractibility at school
Anxiety is one of the most common psychological disorders in school aged children and adolescents (Boyd, Kostanski, Gullone, Ollendick, & Shek, 2000).There is increasing evidence that traditional behavioural cognitive therapies are less successful for social anxiety disorder than for other types of anxiety (Hudson et al., 2015;Norton & Price, 2007).This can have dramatic consequences in education.According to the optimal-level of stimulation model, by promoting a "prevention focus" state, anxiety may have a negative impact on attention and openness towards an unfamiliar teacher, in addition to new teaching material.Supporting this statement, individuals with high math anxiety show higher distractibility (Hopko et al., 2002;Suárez-Pellicioni et al., 2013, 2014).
Considering the contextual stimulation potential could inform the development of future interventions.In fact, for anxious P. Ibáñez de Aldecoa et al.
children, the school environment may present too much distraction (Fisher et al., 2014) and too many stressful situations in which they may feel ridiculed, excluded, or victimized (Pascoe et al., 2020).Future experiments could control explicitly for such variables.For instance, we may hypothesize that anxious children will engage more with a new pedagogical material in a familiar or understimulating environment compared to an unfamiliar or over-stimulating one.Similarly, no studies have yet compared the impact of homeschooling on this specific group of children.
Considering the effects of the social environment can also be informative as the additional levels of stimulation and emotions elicited by the teacher or the school psychologist might play a decisive role on anxious children's engagement.To our knowledge, very few studies have investigated the possible effects of teacher or therapists' characteristics, such as gender, age or ethnicity, on a child engagement.In adults, one study showed that gender-matched patient-therapist dyads reported higher alliances and were more likely to complete treatment (Wintersteen et al., 2005).Racial matching also predicted greater retention but not patient-rated alliance.Age could also have an impact, as demonstrated by the success of certain peer mentoring programs (Sanchez et al., 2006) but to our knowledge, no studies have yet compared the impact of peer mentoring (vs classical mentoring) on anxious children.

Conclusion
The optimal-level of stimulation model suggests that the level of stimulation processed by each individual must be neither too low nor too high.Attentional shifts and exploratory behaviours can be viewed as strategies to adjust the stimulus field in order to stay closer to one's optimum.When the surrounding stimuli are getting too complex, individuals can either focus and explore simpler stimuli or adopt a more global processing style to broaden their mental categories, making the stimuli appear more familiar.When stimuli are too simple, individuals can either look for more complex stimuli or be creative.The need and the ability to explore other stimuli, to adopt a different processing style or to be creative will depend upon the cognitive resources available.These resources can be affected by the contextual stimulation and by other factors such as age, mood and arousability.
Once considered the role of contextual stimulation potential, as well as the diverse factors influencing internal cognitive resources, the optimal-level of stimulation model, as updated in this article, is simple, holistic and coherent with previous literature.It provides an alternative approach which dissolves conceptual inconsistencies and methodological problems related to familiarity and novelty "preferences".Moreover, it fits into and enriches existing theories.By doing so, it opens a new avenue of research for proximate explanations underlying exploratory behaviours and cognitive development, in addition to providing a useful framework for developing future interventions related, for instance, to eating and learning disorders.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Fig. 4. Updated optimal-level of stimulation model including contextual stimulation potential and age.