A shared novelty-seeking basis for creativity and curiosity: Response to the commentators

Abstract

In our target article, we proposed that curiosity and creativity are both manifestations of the same novelty-seeking process. We received 29 commentaries from diverse disciplines that add insights to our initial proposal. These commentaries ultimately expanded and supplemented our model. Here we draw attention to five central practical and theoretical issues that were raised by the commentators: (1) The complex construct of novelty and associated concepts; (2) the underlying subsystems and possible mechanisms; (3) the different pathways and subtypes of curiosity and creativity; (4) creativity and curiosity “in the wild”; (5) the possible link(s) between creativity and curiosity.

R1. Introduction

In our target article, we proposed and substantiated a link between curiosity and creativity and outlined a novelty-seeking model (NSM) that underlies them both. We further suggested that the manifestation of the NSM is governed by one's state of mind (SoM) and demonstrated how this interplay may result in different subtypes of curiosity and creativity. We received 29 intriguing and thoughtful comments from varied research domains, which helped us examine our suggested theoretical model from a broader perspective, while clarifying and alluding to ideas that might have not gotten enough attention in the original article. This enriched the discussion in several important directions.

First, although novelty was first discussed as a relatively monolithic concept, in section R2 we address novelty in more depth and emerges as a multi-faceted concept that is intimately linked with other cognitive dimensions, such as learning, uncertainty, and familiarity. Thus, novelty seems to influence creativity and curiosity in many nuanced ways. Second, several specific subsystems and mechanisms have been raised by the commentaries in reference to our theoretical proposal. These include the division of semantic and episodic memory systems, the default mode network (DMN)/executive control network (ECN) neural systems, dynamic thresholds, and accounts based on metacontrol, mood or developmental changes. We discuss these potential mechanistic accounts in light of the reviewed findings in section R3. Third, in section R4, the expanded discussion following the commentaries resulted in a more parceled framework, with the understanding that factors like usefulness or persistence might pave pathways to different subtypes of curiosity and creativity. Fourth, in section R5, the link between curiosity and creativity is now broadened to include the naturalistic lens outside the lab, with stimulating examples of what drives novelty processes in poetry, art or music appreciation. Lastly, some critical commentaries expressed concern about whether the explored evidence is sufficient for substantiating the NSM. In section R6, we discuss the possible link(s) between curiosity and creativity in more detail. The response discussion below is organized by these five central theoretical and practical issues, addressing the raised suggestions, and highlighting the remaining open questions.

R2. Novelty is a complex construct

The target article proposed the NSM as the fundamental underlying basis for creativity and curiosity. While we referred to novelty in a rather generalized manner, we acknowledge that novelty is heterogeneous and that the kind of information in each case and context determines the novelty type (Kafkas & Montaldi, 2018), in line with findings from novelty detection research.

As suggested by some of the commentaries (e.g., Servais & Bastin; Omigie & Bhattacharya; Holm & Schrater; Becker et al.; Jirout et al.; Singh & Murayama), the distinction between different types of novelty and the circumstances in which they arise should be considered and integrated in the NSM. We take this opportunity to expand the discussion on the subtypes of novelty and their possible different effects on curiosity and creativity.

We embrace Jirout et al.'s suggestion to look at novelty as a continuous concept, which implies that a certain level of novelty optimally drives the NSM. It is indeed reasonable to assume that there is an inverted U-shaped relationship between novelty and both curiosity and creativity. Whereby a moderate level of novelty may result in the optimal performance of the NSM, non-novel or overly novel information is less likely to result in effective recruitment of NSM. Building upon the involvement of dopaminergic activity in novelty detection (e.g., Duszkiewicz, McNamara, Takeuchi, & Genzel, 2019), studies showing a comparable relationship between dopamine levels and different types of cognitive processes (Cools & D'Esposito, 2011), including creative performance (Akbari Chermahini & Hommel, 2012; Boot, Baas, van Gaal, Cools, & De Dreu, 2017; Chermahini & Hommel, 2010), may partially support this hypothesis. For example, Gvirts et al. (2017) demonstrated how the effect of methylphenidate (which increases dopamine levels) on creativity was modulated by participants' level of novelty-seeking: Creative thinking in participants with low baseline levels of novelty-seeking was improved, while creative thinking in participants with high baseline levels of novelty-seeking was impaired.

The idea of novelty as a continuum was also raised by Servais & Bastin who suggested that prior knowledge determines novelty type and further modulates how novelty impacts memory. This relates to the distinction made by Duszkiewicz et al. (2019), and referred to in the target article, between common novelty (i.e., novel stimulation that shares similarities with relevant past experiences) and distinct novelty (i.e., completely new experiences that have minimal connections with past experiences), two ends with a continuous spectrum between them. Servais & Bastin further suggest that these novelty types are likely to tap into novel highly congruent and novel highly incongruent information (respectively). These two extremes are thought to be remembered better than less (in)congruent information (Quent, Greve, & Henson, 2022). This proposal, positioning novelty on a congruent–incongruent spectrum, may explain how novel stimulations may vary in their relative familiarity but still be attended to as novel (e.g., walking on the street and seeing a dog for the first time vs. seeing a specific breed for the first time such as Belgian shepherd).

Incongruency is further related to the surprise effect on memory, which for long was studied in isolation, but might contribute to the distinction between novelty types. Becker et al. call for a clear distinction between novelty, which relates to the unfamiliar, and surprise which relates to the unexpected. The commentators suggest that novelty signals which are triggered by unfamiliar stimuli and situations (either expected or unexpected) and surprise signals, which arise in the face of unexpected stimuli (either familiar or unfamiliar) are mathematically well-defined and may tap into different neural networks (we elaborate on the latter in the next section). We agree, and hope that examining the interplay between novelty and surprise in the future may elucidate different types of novelty. For example, Frank and Kafkas (2021) make a different distinction between expected novelty (stimulus) and unexpected novelty (contextual). Stimulus novelty refers to stimuli that have not been encountered before, and therefore are salient irrespective of expectation. Contextual novelty, on the other hand, relates to the unexpected pairing of familiar and novel inputs in a given context and is driven by expectation violation. These subtypes can be regarded as analogous to the common-distinct novelty continuum, while taking into account the viewer's expectations. Interestingly, according to Quent, Henson, and Greve (2021), the experience of novelty is never “absolute” in the sense that it cannot be defined independently of the observer; rather it is driven by the gap between what individuals expect to experience and what they actually encounter. This view challenges the idea of stimulus novelty that is independent from expectations, revealing the existing gap in our understanding of the different forms of novelty.

In sum, these different accounts highlight the need to unpack the concept of novelty, and delineate how the interplay between expectations and familiarity contribute to novelty. This interplay can be tested by embracing a design in which novelty is operationalized as stimulus novelty, and expectation is determined by the probability of occurrence within a set of stimuli (Frank & Kafkas, 2021). Future studies should further examine how this interplay affects curiosity and creativity. From a broader, theoretical point of view, it would be interesting to consider the mechanism supporting the prioritization of certain types of novelty for different subtypes of curiosity and creativity.

R2.1. Novelty-associated concepts

Beyond novelty, several alternatives that might drive curiosity, creativity, or both have been proposed by the commentaries. These include learning, uncertainty reduction, and familiarity, and we use this opportunity to broaden the discussion to include these related concepts.

R2.2. Uncertainty

The role of uncertainty in both curiosity and creativity is discussed in the target article in light of the SoM framework and is one of the possible links between the two. The question of whether uncertainty reduction not only links curiosity with creativity, but drives curiosity has been raised by several commentators. For example, Holm & Schrater assert that novelty is neither sufficient nor necessary to instill curiosity. They imply that non-novel, but uncertain situations may be enough to induce curiosity, giving as an example a specific case of specific curiosity (the urge to look up for the name of an actor while watching a movie). This example is akin to contextual novelty in which information is not novel, but is incongruent with the context, and as such, this example does not cover the diverse curiosity spectrum. Along similar lines, Singh & Murayama suggest that while creativity might be driven by novelty, curiosity drives people to seek information that reduces uncertainty (Fitzgibbon & Murayama, 2022; van Lieshout, de Lange, & Cools, 2021), rather than to seek the novel. This, too, explains a specific subset of curiosity-related behaviors which are directed towards the aim of filling an information gap, and does not account for all types of curiosity. Similarly, Jirout et al. claim that resolving uncertainty may be a better underlying basis for both curiosity and creativity than novelty-seeking. As suggested in the target article, curiosity may help transform situations of uncertainty from being experienced as threatening to providing a fruitful ground for the generation of novel ideas. This is quite similar to the commentators' approach explained elsewhere (Evans & Jirout, 2023), suggesting that one is curious when uncertainty is identified, and this uncertainty is responded to in a creative act.

These observations provide a clear example of how curiosity, and potentially subsequent creativity, is driven by the need to fill a gap. However, as clearly stated in the target article, not all creative acts or curious behaviors are driven by uncertainty reduction. Following Litman's (2008) distinction, we stated that the interest type of curiosity relates to the anticipated pleasure of new discoveries without benefiting from uncertainty reduction (e.g., curiosity about the movement of the clouds while looking at the sky), and only the deprivation type of curiosity is concerned with uncertainty reduction (e.g., understanding how an electrical device works so we will be able to fix it). Thus, we emphasize here again that while uncertainty seems to be linked with both curiosity and creativity, it does not necessarily drive all forms of curiosity and creativity. It is worth noting that in the target article we proposed that SoM determines which type of curiosity would be likely to be evoked, differentiating between deprivation and interest and clustering each with their governed SoM (and the associated subtypes of creativity). Further exploring how SoM, uncertainty, and novelty interact will be of pivotal importance for better understanding the contexts in which motivations for uncertainty reduction motivations shape novelty-seeking behaviors.

R2.3. Familiarity

People are often curious about the things they are familiar with. This is an excellent observation raised by Singh & Murayama and supported by studies showing that participants choose to learn about subjects they are already familiar with (Alexander, Kulikowich, & Schulze, 1994; Fastrich & Murayama, 2020; Tobias, 1994). While at first these findings might seem to contradict novelty-seeking, they can be explained by the common-distinct novelty spectrum discussed above. We can find novel information about topics that we are already knowledgeable about (or fill knowledge gaps, using the commentators' terminology). In a way, seeking novel information within a familiar context might be a formula for an optimal novelty level, which is not overly costly and at the same time highly learnable. For example, art lovers might get more curious about the opening of a new exhibition than gaining knowledge about other random subjects that they are not familiar with, similar to Tarantino's fans who will curiously anticipate his new film premiere while showing less curiosity about the release of a film by an unfamiliar director. The commentators claim that this knowledge gap (the exhibition or the film in this case) is likely to provide a bigger marginal gain to the audience in terms of understanding the topic; but they are nonetheless encoded as novel. In other words, familiar is not necessarily the opposite of novel. This is supported by novelty-detection theories according to which novelty and familiarity offer distinct contributions to recognition memory decisions, showing that their signals originate from non-overlapping brain regions (Kafkas & Montaldi, 2018).

R2.4. Learning

Omigie & Bhattacharya suggest a qualification of “novelty-seeking” as the pursuit of learnable novel information. They state that according to progress learning theories (e.g., Oudeyer, Kaplan, & Hafner, 2007), “curiosity is not solely driven by novelty but is precipitated by heightened rates of learning new information.” Along similar lines, Liquin & Lombrozo further argue that expected learning and utility are primary to novelty, such that explicit signals of expected learning and utility are primary to triggering curiosity, compared with implied signals such as novel cues.

Indeed, learning plays a key role in driving curiosity, and we emphasize its importance in the NSM. We proposed that when we commit to the combination we generated in previous stages, it will result not only in consolidation, but with further learning and elaboration (p. 34). In other words, we agree that stimuli that have no learning potential will not be attended to in the first place or at most will fall in one of the subsequent stages of the model. That said, learning is not sufficient, it is the desired outcome but not necessarily the primary drive. Curiosity is driven by signals that are ultimately novel and meaningful. Similar to the interplay between usefulness and originality in creativity, an optimal balance between novelty and meaningfulness will result in consolidation and learning.

Omigie & Bhattacharya further postulate that curiosity limits wasting resources on irrelevant or overly simple content on the one hand, and on content that is too complex for our current understanding on the other. In other words, curiosity directs the agent toward moderately novel and learnable information. This is also explained by our model, and in line with the inverted U-shaped relationship mentioned above, whereby while we are attracted by novelty, not every novel stimulation will necessarily trigger our curiosity.

This discussion brings up the question of whether there is such a thing as too novel? For example, Gustafsson et al. claim that “an excess of novelty may be overstimulating, perceived as noise, and not necessarily evaluated positively.” In our model, we proposed that the level of novelty is assessed during the multi-staged process of the NSM, and information that is too novel or that is insufficiently novel is more likely to be ignored, or later be ruled out. The same principle is true for creativity: Ideas that are not sufficiently original on the one hand, or overly bizarre on the other hand, will be ruled out and will not come to light. This is in line with the commentators who suggested reframing the need for novelty as a need for an optimal level of arousal (in which exploratory behaviors are used to maintain the arousal level). Both cases highlight the need for moderate-level novelty as a drive for creativity and curiosity. That said, future studies in both the curiosity and creativity realms are needed to examine this premise more thoroughly.

Furthermore, the notion that novelty-seeking can be rather costly, was raised by several commentators (e.g., Gustafsson et al.; Omigie & Bhattacharya), and we indeed agree that too much novelty-seeking can be unadaptive. By further addressing the subtypes of novelty, we were able to better allude to the costs of novelty in our model. As pointed out by Zhou & Berenstein, the desire to explore and learn novel information must thus be balanced in order to avoid an overflow of information. As indicated by the commentators, one way to manage this is to use heuristics that locally track what has and has not been seen before. By contrast, familiarity may be used to reduce uncertainty and acquire missing information. While the latter can be seen as a form of perseverative information-seeking and has been associated with deprivation curiosity, the former might be linked to diversive curiosity. This is in line with the SoM framework and with the suggested matrix that we composed, according to which hyper-exploration will result in an overflow of information while hyper-exploitation will result in a stand-still perseverative state. In either case, if not balanced, both can result in dysfunction. Extreme preference for familiarity is prevalent in depression and anxiety, while attraction to excessive novelty might be associated with hypomania and psychosis (e.g., Baas, Nijstad, Koen, Boot, & de Dreu, 2020).

On a related note, Kashdan et al. point to the possible costs creativity and curiosity hold (metabolic, psychological, social), which when maximized may produce overstimulation due to vigilance and may result in dysfunction. Similar to an optimal level of novelty or arousal, optimal level of curiosity and creativity balances their costs and benefits based on contextual demands. In line with the matrix suggested in the target article and with the inverted U-shaped function, too little or too much creativity or curiosity may result in negative outcomes to the point of psychopathology.

To conclude, in the target article, we referred to novelty in a very general manner to be able to construct a unified and parsimonious model that explains the commonalities between different novelty-seeking behaviors (e.g., curiosity and creativity). Based on the commentaries cited above, here we decomposed novelty, taking into account potent features (expectations and familiarity), and examined the different subtypes of novelty resulting from the interplay between these concepts. By doing so we show that although several variables interact in the process of curiosity and creativity, as alluded to in the target article, novelty-seeking best explains the underlying basis of human curiosity and creativity. Interestingly, our assumption is supported by recent computational evidence showing that when exploring complex environments, novelty-seeking is the most probable model of human behavior, outperforming seeking information-gain or surprise (Modirshanechi, Becker, Brea, & Gerstner, 2023). We encourage future research to empirically test whether different types of novelty commonly guide curiosity and creativity.

R3. Subsystems and possible mechanisms

In the target article, we pointed out possible neural systems that subserve the NSM. However, as neural evidence is still lacking, how novelty is processed in the human brain, and the specific neurochemical systems involved are still elusive. We therefore embraced a more theoretical approach and avoided overly speculative mechanistic explanations, while noting that the networks and mechanisms suggested are not solely involved in curiosity and creativity, a concern that was raised by some of the commentaries. This is a challenge neuroscience often faces when attempting to assign a unique function to a specific cortical region or network. A more mechanistic approach is certainly needed to advance our understanding of the possible link between curiosity and creativity, which will be increasingly possible with more research. Our model provides a basis for future investigations to suggest the involvement of potential mechanisms, as some of the commentaries also attempted to do (e.g., Servais & Bastin; Chiou et al.; Zeitlen et al.; Faber & de Rooij; Prasad & Hommel). These are important additions, and we discuss them in more detail below.

R3.1. The interplay between episodic and semantic memory

A great part of the discussion in the target article on the role of memory in creativity and curiosity was dedicated to semantic memory (although the role of episodic memory was briefly alluded to, e.g., Duszkiewicz et al., 2019; Madore, Addis, & Schacter, 2015, 2019). This is mainly due to the fact that the bulk of evidence reviewed by us investigated semantic memory as the main system involved in both. However, we acknowledge that understanding the role of episodic memory in creativity and curiosity will certainly shed more light on their common basis.

Interestingly, the interplay between these two memory systems is closely related to novelty detection and may tap into the possible subtypes discussed above. Servais & Bastin point out that the distinct subtypes of novelty lead to memory representations of different nature. In common novelty, new information that is congruent with prior knowledge will be combined with prior semantic knowledge to fill an existing gap within the associative semantic network. On the other hand, distinct novelty is thought to form episodic, standalone memories. The commentators further hypothesize that the two ends of the U-shape are linked to different memory systems: While congruent information is supposed to be stored as semantic representations, incongruent information is thought to induce episodic memories. Although appealing, this assumption may be true in limited cases, as encoding of episodic experiences can be schema-congruent at times. Moreover, the view of semantic and episodic memory as functionally distinct is gradually replaced by seeing them as complementary and interrelated memory systems (Benedek, Beaty, Schacter, & Kenett, 2023). Future studies should consider in more depth their complementary contribution to both creativity and curiosity. Given the view of interdependency between semantic and episodic memories, such investigations are likely to result in a more intertwined mechanistic structure than a parallel dual-paths model.

R3.2. Subsystems of DMN/ECN

In accordance with the involvement of episodic and semantic memory, Choui et al. suggest that both the DMN and ECN are functionally fractionated into subnetworks, which potentially enables distinct facets of creativity (and curiosity). According to their proposal, there is a dissociation between a “semantically oriented subnetwork,” which is associated with semantic memories, evaluative cognition, and convergent thinking, and an “episodically oriented subnetwork,” which is more associated with episodic memories, free association, simulating hypothetical scenarios, and divergent thinking. Their suggestion accords with the distinct roles of memory systems discussed above and provides an initial framework for the differences between creativity induced by semantic and episodic memories, and the distinct subsystems involved in each case.

Another important issue related to the heterogeneity of those networks was raised by Benedek who commented about the interaction of the DMN and ECN with various processes–stages in the NSM. For example, DMN structures also contribute to idea evaluation (Benedek et al., 2023), and cognitive control is also involved in idea generation (Benedek & Jauk, 2019). While we only schematically associate neural networks with specific stages along the model, based on previous evidence, we agree that their contribution is likely to go beyond a single, specific stage. This may be valid also for Prasad & Hommel's assertion that attention is implicated in all four stages and not only in the affinity stage, or that of Gabora et al., who interpreted the model as composed by distinct rather than interweaved phases. In the target article, we alluded to these dynamic interplays between the networks and between the stages (such as the cyclic motion between evaluative and generative processes), but we agree that the dynamic sub-interactions within the systems as postulated by Chiou et al. may promote a more refined understanding of the mechanisms involved throughout the process.

Considering these dynamic sub-interactions may further explain the potentially differential neurochemical effects on the distinct subtypes of creativity and curiosity. Although beyond the scope of the target article, this is a valid assumption that should be tested, as mentioned by Baas et al., suggesting that these differential effects ultimately feed into creative thinking and doing (Beversdorf, 2019; De Dreu, Nijstad, & Baas, 2024). Other than the involvement of dopamine, which we describe in the target article, surges in norepinephrine among others may be involved in novelty as we discuss next.

R3.3. The role of the noradrenergic system/dynamic threshold

In light of the interesting commentary by Faber & de Rooij, we discuss here the possible role of the noradrenergic system in the NSM. Based on the adaptive gain theory (Aston-Jones & Cohen, 2005), which proposes that locus coeruleus norepinephrine (LC-NE) serves to adjudicate the trade-off between exploration and exploitation, the commentators suggest that the activation threshold, described in our model, may be adaptive to contextual demands. For example, the threshold may be more selective (higher) when one is stuck, looking for a tailored solution for a specific problem, and this threshold is adaptively lowered during incubation, allowing for more ideas to attract our attention so progress can be made.

The adaptive gain theory is of great relevance, originally suggesting that exploitatory mode is driven by phasic norepinephrine (NE) for prioritizing related information and tonic NE is associated with an exploratory mode promoting search for other alternative behaviors. Upshift in tonic NE enhances functional connectivity in relevant networks and reduces phasic responses to the extent that attentional decoupling occurs, which facilitates a more exploratory mode. As further highlighted by the commentators, while moderately heightened tonic NE, which increases sensitivity to novel information, may facilitate creativity through defocusing; phasic NE enhances response to salient events and thus helps in the selection of the ideas that will cross the threshold. These dynamics shed more light on the possible role of NE in novelty-seeking and are in line with the readiness potential, described in the target article as the driving mechanism for novelty detection (see Broday-Dvir & Malach, 2021, for detailed description). During this slow uprising phase of spontaneous fluctuations in cortico-hippocampal circuits, a spontaneous mental event can emerge. This is followed by low-level activation spread in relevant networks generated by any new content (Noy et al., 2015) and ends with dopamine release in the hippocampus. This bodes well with the adaptive gain theory that proposes that these systems work in synergy: The LC-NE system regulates the balance between exploitation and exploration, and the new knowledge is implemented and rewarded by the DA system (Aston-Jones & Cohen, 2005).

Faber & de Rooij further suggest that the adaptive gain theory may explain how saliency is affected by changing environmental demands. As an example, the commentators postulate that in response to an impasse, this upshift in tonic NE might reduce the bias to previous neural activity that has led to the impasse and lower the novelty threshold accordingly, raising the chances for more spontaneous ideas to become salient. However, as a lowered threshold might come at the cost of accuracy, when overcoming the impasse, the threshold is adaptively raised, through more deliberate modes of creative thought. Although the commentators propose how incubation “resets” the threshold through attentional decoupling, allowing for spontaneous ideas to emerge from subsequent mind wandering; this mechanism may possibly explain the interplay between spontaneous and deliberate processes in a broader sense. Our suggestion that the activation phase is mediated by the salience network is supported by the adaptive gain theory, proposing that the anterior cingulate cortex (ACC), a key region in the salience network (SN), regulates the transitions between phasic and tonic NE. While future research is yet to determine the role of the noradrenergic system in curiosity and creativity, understanding the adaptivity of the activation threshold to contextual constraints is of great importance.

R3.4. Metacontrol

The expanded discussion here about the possible neurochemical mechanisms involved in balancing the trade-off between exploration and exploitation may provide a mechanistic, cognitive-control related element that is lacking in the target article, as suggested by Prasad & Hommel. The commentators highlight the importance of cognitive control in balancing opposing forces or control strategies by switching from top-down to bottom-up states of processing. This idea shares significant commonalities with the description of exploration as derived from bottom-up processing and exploitation from top-down control, as shown in the target article and explained originally and in detail by the SoM framework (Herz, Baror, & Bar, 2020). We hope that the expanded discussion here about the possible neurochemical mechanisms involved in the balancing trade-off between exploration and exploitation adds clarity.

Furthermore, in line with the contributions of the metacontrol model (Hommel, 2015) or of reinforcement models such as those of Becker et al., we encourage future studies to quantify the respective contributions of exploration and exploitation to the NSM and test which mechanisms regulate the trade-off between exploratory and exploitatory states. We also accept Arbib's notion, by which the SoM framework didn't elaborate how exploration and exploitation relate to the DMN, SN, or ECN. With this gap, it would be important that future investigations examine how interactions within DMN–SN–ECN are influenced by one's SoM on the exploration–exploitation continuum, and whether in accordance, these mechanistic interactions are represented in the different subtypes of curiosity and creativity.

Lastly, Baas et al. point out that the role of neural oscillations captured by EEG is ignored in the NSM. Although there is a growing body of evidence regarding the role of alpha power in creativity, as the commentators propose, EEG studies of curiosity are lacking (but see Appriou et al., 2020). We agree with the commentators that a combination of different neuroscience methods will ultimately reveal the unified nature of curiosity and creativity and thus invite future studies to examine their neural basis via varied methods.

R3.5. Mood

Some commentaries emphasized the possibility that mood regulation might serve as a shared mechanism to promote creativity and curiosity. Although only briefly mentioned in the target article, according to the SoM framework (Herz et al., 2020), affect is a major pillar of SoM, together with dimensions such as perception, attention, thinking, and openness to experience (i.e., the exploration–exploitation continuum). Whereas broad SoM is associated with positive mood, with creativity and with exploratory behavior, narrow SoM is associated with negative mood, a constricted thinking pattern and exploitatory behavior.

Zeitlen et al. support the reciprocal connection between curiosity and creativity by suggesting how both are commonly driven by the desire to enhance positive moods and/or reduce negative moods. In line with their proposal and with the SoM framework, specific curiosity and convergent thinking are associated with negative mood reduction, and diverse curiosity and divergent thinking with enhancement of positive mood (see also Bar, 2009). The suggestion that curiosity and creativity are purposeful vehicles for mood regulation is an appealing avenue for future research.

Zeitlen et al. further suggest that the link between curiosity and creativity goes beyond mood regulation, in which mood can promote information-seeking by increasing the perceived value of information. In the same vein, Gustafsson et al. suggest that mood changes one's cognitive resources and arousal levels of the stimuli met in a given context. A stimulus must have a potential positive reward to trigger the intrinsic motivation to explore. The reward is context-dependent such that a very simple stimulus, usually perceived as boring, could be positively rewarding if one is under-stimulated. It is important to consider, however, cases of morbid curiosity (as raised by Baas et al.), in which we may be curious about negative information. This negative information, however, may be assigned with positive value, driving one's curiosity and will result in new rewarding knowledge. In the same manner, this may explain cases of dark or malevolent creativity, in which novel ideas aimed at damaging others (Perchtold-Stefan, Fink, Rominger, & Papousek, 2022).

Furthermore, Zeitlen et al. states that the unified process underlying creativity and curiosity is flexibly affected by mood: “positive mood typically promotes an exploratory SoM, fostering creativity and diversive curiosity; but when a local style is dominant in one's SoM, then negative mood promotes an exploratory SoM and positive mood might promote an exploitatory SoM, fostering specific curiosity.” This is perfectly echoing our premise in the target article, and in more detail in Herz et al. (2020) and Bar (2009). Indeed, mood reinforces processing style, as embedded by SoM, and is directly influenced by the current context. We further suggest that both exploitatory and exploratory SoMs may result in creativity/curiosity (novelty-seeking), but of different types. This framework adds a more nuanced approach to understanding motivational and contextual conditions in which curiosity and creativity may flourish.

R3.6. Developmental perspective

The NSM could be applied across the lifespan, as suggested by Vaisarova & Lucca. The commentators point to evidence showing that children do not show greater ideational fluency, as might be predicted by our model. As an example, they assert that there is a negative correlation between fluency and executive functions in kids, as opposed to adults. The latter is of interest as the commentators further suggest that there might be a qualitative change in top-down processing across age. In line with the cyclic motion between generative and evaluative processes and the evidence that regions in the ECN are activated during generation of ideas, it is possible that the interplay between the processes changes with age. Indeed, studies demonstrate that adults show stronger functional connectivity between and within the DMN and ECN up to adulthood (Fair et al., 2008; Sherman et al., 2014; Uddin, Supekar, Ryali, & Menon, 2011), and more specifically an increased SN influence, which guides the switching between those networks, was found across development (Uddin et al., 2011). Similar findings were evident during a divergent thinking task, where older adults showed stronger functional coupling between the DMN and ECN compared with young adults, implying that greater default-executive functional coupling occurs with age (Adnan, Beaty, Silvia, Spreng, & Turner, 2019). These dynamics might be affected by the social context as further suggested by the commentators and supported by studies who found that cultural (Ivancovsky, Kleinmintz, Lee, Kurman, & Shamay-Tsoory, 2018), school-experience (Duval et al., 2023), and expertise (Kleinmintz, Goldstein, Mayseless, Abecasis, & Shamay-Tsoory, 2014) may modulate these brain networks in creative thinking.

Similarly, children may be less successful in governing the trade-off between exploration–exploitation and switch between these strategies throughout the process. While children are exploratory in nature, utilizing the advantages of exploitatory behavior may increase across development with the gradual maturation of their prefrontal cortex and the acquisition of experience. Furthermore, adults' exploratory behaviors, such as curiosity and creativity, may inherently require exploitation abilities (e.g., inventing a useful tool; see Neldner et al., 2019) and rely less on “pure” exploration as children (see Gopnik, 2020, for a more elaborated discussion). It is thus possible that in accordance with the discussed subtypes of novelty, children will experience more stimulus novelty, as indeed more experiences are truly novel for them, and will gradually switch to contextual novelty in adulthood.

Considering developmental changes in the concept of novelty would be an interesting perspective for future investigations. For example, a recent developmental study found that whereas adolescents and adults demonstrate attenuated uncertainty aversion for more novel choice options, children's choices were not influenced by reward uncertainty when choosing options that entail greater novelty (Nussenbaum et al., 2023). Developmental accounts are therefore of great importance for a more comprehensive understanding of novelty-seeking, curiosity, and creativity.

R4. Different pathways lead to different CC subtypes

In the target article, we proposed a generalized model to describe and connect curiosity and creativity. After substantiating the model, we then described the subtypes of both curiosity and creativity and suggested possible interactions between these subtypes and the SoM framework, resulting in two endpoints of a continuum, one more associated with exploration and the other with exploitation. It is possible that those clustered subtypes are interconnected by different cognitive pathways, associated with distinct neural subsystems, as implied by some of the commentaries. Alternatively, alterations between these two yet-hypothetical pathways throughout the NSM may be necessary to optimize performance based on context and the available resources. In line with the SoM framework, we suggested that these shifts throughout the NSM are determined by the balance between top-down (TD) and bottom-up (BU) processing (Herz et al., 2020). Future research is yet to determine whether these clusters represent two separate cortical infrastructures or are associated with the same infrastructure, but with varying weights that give an advantage to TD or BD processing. As empirical evidence that links curiosity and creativity is still lacking, our model sets the ground for such investigations.

R4.1. The role of usefulness

Utility is an important aspect of every human behavior, and its role in curiosity and creativity has been rightfully emphasized by the commentators (Acar & Fuchs; Liquin & Lombrozo; Litovsky et al.; Runco). Considering that creativity requires both novelty and usefulness (Runco & Acar, 2012; Sternberg & Lubart, 1996), and that curiosity is rewarded by new knowledge (i.e., learning) and novel experiences (Kashdan & Silvia, 2009; Litman & Jimerson, 2004), utility potentially plays an important part in both. As further suggested in the target article, novelty should be accompanied by usefulness for avoiding bizarre or useless/esoteric ideas or information. This balance is achieved through iterations between the affinity phase, where novelty is prioritized, and the evaluation phase in which other valuation components are taken into consideration, such as usefulness and appropriateness. The balance between the two is further governed by SoM; in exploratory SoM, we prioritize novelty and diversity, and in exploitatory SoM, we lean toward immediate utility (e.g., filling a knowledge gap, finding a solution to an existing problem/elaborating on an existing idea). Indeed, utility has been suggested to influence exploration–exploitation tendencies: When high utility is predicted by prior events, exploitation is enhanced, and if low utility is predicted, exploration of novelty emerges (Aston-Jones & Cohen, 2005; Cohen, McClure, & Yu, 2007).

As research on usefulness is somewhat of a later contribution to the work on creativity, the relationship between usefulness and originality is disputed (Harvey & Berry, 2023). Acar & Fuchs underscore the distinction between novelty and usefulness, suggesting that these two components might be independent of each other. Runco points to the possibility that novelty and usefulness are simultaneously processed. While Acar & Fuchs are concerned by the (over) interdependency between usefulness and originality suggested in the target article, and Runco's concern is the complete opposite, holding that usefulness and originality are presented in the model in clear separation. These conflicted readings clearly reflect the debate in literature. Relying on dual models of creativity (e.g., Basadur, 1995; Sowden, Pringle, & Gabora, 2015), we proposed that novelty and usefulness are balanced throughout a cyclic motion and iterations between generative and evaluative processes, implying their interrelated nature. Although we describe them as sequential in our model, whereby affinity is triggered by novelty and only then assessed for its usefulness, we do not see them as isolated processes, as implied by Runco, but rather as two critical parts within a more comprehensive process.

Runco makes an important observation that usefulness changes from one domain to the other. In the same vein, social and cultural factors may influence the balance between originality and usefulness in creativity, favoring one or the other (e.g., Erez & Nouri, 2010; Ivancovsky, Shamay-Tsoory, Lee, Morio, & Kurman, 2021). This is in line with Vaisarova & Lucca's developmental perspective suggesting that “social cues may draw individuals’ attention to certain aspects of their environment, as well as shaping their goals and evaluation criteria. Standards for the usefulness of an idea, for instance, might be higher in a context where it will be used by others.”

Liquin & Lombrozo argue that in curiosity, utility is primary to novelty and that curiosity is sensitive to novelty mainly because it signals that useful learning is likely. They support their assertion by demonstrating how sensitivity to utility triggers is likely to produce “optimal” patterns of curiosity: High curiosity when useful learning is most likely and most rapid, and low curiosity when useful learning is least likely and least rapid (Liquin, Callaway, & Lombrozo, 2020, 2021; Poli, Serino, Mars, & Hunnius, 2020). This pattern of results can similarly reflect the balance between novelty and usefulness, explaining why too-novel information that completely lacks usefulness is less likely to be consolidated. While the priority of novelty and usefulness in curiosity may be debatable, considering the interplay between them, as suggested in the target article and discussed in further detail here may hold a key function.

On the contrary to the commentaries discussed above, Litovsky et al. criticize the inclusion of utility whatsoever, stating that the model fails to explain why non-instrumental information is attended to (e.g., gossip), or why valuable information is sometimes ignored. Acar & Fuchs further suggest that novelty-seeking might be unrelated to usefulness, proposing that curiosity may lead to prioritizing novelty over usefulness, referring to studies that found that curiosity and interest are crucial for the initial phase of the creative process (Amabile, 1996), and that individuals with a strong motivation to acquire new knowledge generate less useful solutions to innovation problems (Acar, 2019). These examples are in line with our suggestion that the subtypes of curiosity and creativity may be interconnected in a distinct way: While originality is linked to interest or diversive curiosity and divergent thinking, usefulness may be linked to specific curiosity and convergent thinking.

Interestingly, Dubey, Griffiths, and Lombrozo (2022) suggest that curiosity may arise for information that may seem initially unimportant – if people come to appreciate its usefulness. Litovsky et al. claim that other theories, such as the information gap theory (Loewenstein, 1994), can predict when curiosity will be piqued, but link this claim uniquely to specific/deprivation curiosity, thus, do not explain other forms of curiosity. Furthermore, studies show that uncertainty reduction, associated with this type of curiosity, is also guided by utility. As Liquin & Lombrazo suggest, people aren't curious about all uncertain stimuli (i.e., information gaps), but specifically those likely to be useful in the future (Dubey & Griffiths, 2020; Dubey et al., 2022). If this is the case, the question that Litovsky et al. raise still remains. Information gaps are part of our enriched environment, as we cannot attend them all, some are prioritized based on their novelty, utility, and individual differences.

Why people avoid useful information and choose not to resolve their information gaps, even if it is optimal to do so, remains a question for future research. As Horton & Mason acknowledge, we understand relatively little about the circumstances that prompt an individual to innovate or to attend to novel information rather than settle for a standard or routinized approach. Our model sets the ground for identifying the conditions under which people self-initiate curious or creative pursuits.

R4.2. Persistence versus flexibility

The SoM framework discussed in the target article highlights the tension between exploration and exploitation, and their corresponding function in novelty-seeking behaviors. This tradeoff is reiterated in the commentary made by Baas et al., who elaborate on two cognitive pathways (De Dreu, Baas, & Nijstad, 2008): Flexibility in which original ideas are generated by switching between broad cognitive categories, and persistence in which equally original ideas are generated through a systematic exploration of a semantic category in-depth, evaluating and discarding more readily available ideas. These parallel the broad and narrow SoMs introduced in Herz et al., 2020. While the flexibility pathway is linked with curiosity, novelty-seeking, openness to experience, positive mood, and desirable outcomes, persistence is associated with working memory capacity, negative mood, and threatening circumstances and may result in morbid curiosity and dark creativity. Bringing these ideas together, we identify that flexibility is key to exploration and persistence is analogous to exploitation. These similarities are also acknowledged by Benedek, who ingeniously listed most complementary or opposing concepts mentioned by us and others, using different labels for largely the same thing. We hope that using converged terminology may advance future interdisciplinary discourse.

As further mentioned by the commentators, the balance between flexibility and persistence helps avoiding distractibility and bizarre ideas on the one hand (too much flexibility) or rigidity on the other (too much persistence), similar to the dynamic interplay between exploration–exploitation and cognitive control suggested by us. This interplay may also be addressed by the usefulness–originality continuum perspective. As succinctly summarized by Prasad & Hommel, behavior requires balancing two extremes, and integrating the situational constraints with personal goals. We believe that similar to originality and usefulness, persistence and flexibility mirrors the exploration–exploitation balance, and although it has been claimed that the exploration–exploitation distinction does not gain added value from the SoM framework (Arbib), we do believe that clustering the extreme ends along different dimensions under an overarching framework is beneficial in explaining how all those continuums change in tandem, and how various forms of creativity and curiosity interact. In this sense, Moldoveanu's exploratory exploitation and exploitative exploration concepts are intriguing. It is possible that exploration and exploitation are weaved in a more complex or intermixed manner, but more evidence that pinpoints these subtypes of connections is needed to support such hypotheses.

Interestingly, Horton & Mason indeed call for a deeper dive into how various forms of curiosity and creativity are related. They suggest that considering the multiple pathways in which curiosity might affect creativity may offer a more comprehensive approach. For example, they propose a nuanced and elegant distinction between depth-curiosity and breadth-curiosity to clarify their corresponding contribution to creativity. Depth-curiosity is analogous to the specific/deprivation type, and breadth-curiosity is analogous to the interest/diversive type. These semantic labels used by the commentators may help portray the distinction between these subtypes more accurately. Second, adopting these labels may better demonstrate the relationship with the different creativity subtypes and with the SoM framework as depicted in our model, which may further illuminate how exploration and exploitation interact. Horton & Mason add support by suggesting that depth curiosity may lead to the development of expertise and enhanced problem-solving skills (Harrison, Sluss, & Ashforth, 2011; Lydon-Staley, Zhou, Blevins, Zurn, & Bassett, 2021; Zhou, Xiao, & Zhang, 2020), while breadth curiosity may be instrumental in making remote associations. This is in line with our suggestion that diverse curiosity is closely related to divergent thinking, while specific curiosity is associated with convergent thinking. Depth and breadth curiosity further tap into the persistence (in-depth exploration) and flexibility (switching between broad categories) pathways. Although using slightly different terminology in our model, we acknowledge the great contribution of the pathways approach to the future establishment of a unified but nuanced model of curiosity and creativity.

R5. Creativity and curiosity “in the wild”

Although beyond the scope of the target article, understanding real-life curiosity and creativity is of importance but typically less investigated. According to Omigie & Bhattacharya, creativity and curiosity in areas such as art, music, and poetry is driven by learnable novelty. As an example, they cite Bianco, Ptasczynski, and Omigie, 2020; Cheung et al., 2019; Omigie and Ricci, 2023, demonstrating that learning a new musical style (i.e., the outcome of being curious) significantly predicted success in composing creatively in that new style. They further cite as an example Zioga, Harrison, Pearce, Bhattacharya, and Di Bernardi Luft (2020) who found that musical compositions with moderate, rather than high levels of novelty (i.e., music intervals not previously heard), were judged by listeners as being the most creative. This interesting finding demonstrates how our suggested inverted U-shaped function of optimal novelty levels extrapolates to real-life creativity.

Another example that Omigie & Bhattacharya give to support the learning–creativity link is that aesthetic appeal is a stronger predictor than surprise of how creative poems are judged to be (Chaudhuri, Dooley, Johnson, Beaty, & Bhattacharya, 2023). According to the NSM, the novel stimulation is evaluated based on different criteria including the relevance of the content to the context. Aesthetic appeal may relate to the usefulness component in art rather than to mere learning–creativity associations as suggested by the commentators. This is also in accordance with Runco's suggestion that usefulness in art is portrayed by aesthetic usefulness, rather than practical function or norms.

We join the commentators in stressing the need for research on curiosity and creativity to go out of the lab and investigate real-life creative behaviors. Under such circumstances, a broader set of motivations may come into play, as well as one's commitment to the creative outcome, as is the case for players and artists. This insight is reflected in a commentary by Pagnini et al., who propose that mindfulness is a necessary condition for both curiosity (novelty-seeking) and creativity (novelty-producing), indicating that both factors are typically highly correlated. The commentators interestingly suggest that rigid thinking patterns when mindless prevent curiosity and creativity. When mindful, people are open to novelty, pay attention to the variability of the experience, and adopt multiple experiences, which all facilitate curiosity and creativity. One's level of intention and mindfulness is likely to change in out-of-lab settings, and this element should be taken into consideration.

Moreover, while we mainly focused our discussion on lab-based experiments of creativity, which heavily rely on semantic or linguistic knowledge and problem solving, real-life creativity, such as sculpting, design, or cooking, involves more domains than that. An interesting question that applies to both curiosity and creativity, and was briefly mentioned when discussing familiarity, is the question of domain-specificity. Artistic creativity involves visual, procedural, improvisational skills among others. Yet are curiosity and creativity domain-specific or domain-general? Arbib depicts in his commentary the case of architecture and design creativity and describes in similar terms how architects navigate their own mind (or locometric maps) to turn scripts into buildings. It is likely that mental simulation, guided by the relevant knowledge structure, is one such common process. Future studies are yet to determine whether creative ideation shares similar memory processes across domains (see Benedek et al., 2023).

In the context of “real-life,” Servais & Bastin suggests that the NSM could also address the case of brain-damaged patients with memory problems and the fact that they present decreased creativity (Duff, Kurczek, Rubin, Cohen, & Tranel, 2013). Interestingly, as mentioned in the target article, several lesion studies and case studies on FTD patients (fronto-temporal dementia) among others showed enhanced artistic creativity abilities (see Geser et al., 2021, for a review). These findings may be also relevant for the domain-specific/general discussion above as patients might show dissociation among enhanced and decreased creative abilities. To the best of our knowledge, there are no studies investigating curiosity among brain-damaged patients, but this may be an important future research line to shed more light on the neural underpinnings of both. For example, does curiosity remain intact among patients with memory problems? Would curiosity inductions help mitigate impairments to memory functioning? As some evidence shows that curiosity (and novelty-seeking in general) may have lifelong benefits for memory abilities by affecting hippocampal performance and structure (see Sakai et al., 2018, for a review), it may be the case that a form of curiosity-training may aid memory deficits.

R6. Linking curiosity and creativity

Finally, questions regarding the nature of the mutual connections between curiosity and creativity were raised by some of the commentaries (e.g., Becker & Cabeza; Zhou & Bornstein; Karwowski & Zielińska; Grüning & Krueger; Singh & Murayama). As Benedek points out, in the target article, we performed a “mental factor analysis” on the correlates of creativity and curiosity to extract the factor explaining the most shared variance. We are happy to witness a growing body of research linking curiosity and creativity published since we first submitted our target article (e.g., Evans & Jirout, 2023; Kenett, Humphries, & Chatterjee, 2023; Li, Emin, Zhou, Zhang, & Hu, 2023). In the discussion below, we focus on the issue of causality and on what could be claimed and safely suggested with the current state of research.

Singh & Murayama claim that curiosity and creativity differ in their basic motivation: While people are curious about knowledge gaps between closely related concepts, creativity requires the ability to connect semantically distant concepts. As stated by the commentators, according to the knowledge network theory “…people often feel curious when they perceive the potential for adding new edges between semantically close concept nodes,” while “creativity motivates people to take substantial semantic leaps away from the current stimuli and connect the pieces of information that were far apart.” Although the commentators provide it as an example of the differences between the two, we believe that this aspect is what brings them together. The more interlinked nodes in one's associative network, the easier it would be to make those leaps and connect remote concepts. As suggested in the target article, this notion that “consolidation of interlinks in memory, as curiosity seems to promote, would enable connecting nodes in a novel manner and the shortening of path lengths in the network” (p. 14) is supported by network-science studies showing that increased interconnectivity between concepts tend to characterize associative networks of creative thinkers (Benedek et al., 2017; Gray et al., 2019; Kenett, Anaki, & Faust, 2014).

Interestingly, as proposed by Zhou & Bornstein, curiosity and creativity may both be linked with a form of mental navigation through complex knowledge structures that span diverse spaces. That knowledge may be more modular and compressible, allowing for the grouped representation of a more diverse chain of actions. According to their computational approach, while curiosity expands one's knowledge, creativity compresses existing knowledge by chunking or recombining information, thus managing the cost of increased complexity. These computational processes can be seen in the same vein as expansion of interlinks and shortened paths as described above, supported by studies that found that associative networks of high creative individuals are more condensed, as more concepts are clustered and less rigid than low creative individuals (Kenett et al., 2014; Kenett & Austerweil, 2016; Li, Kenett, Hu, & Beaty, 2021). The commentators mention that the DMN contains the most compressed neural activity that has been measured, which again supports this notion.

Becker & Cabeza suggest that curiosity and creativity share the same underlying computational principle of prediction error minimization. While curiosity corresponds to an expected gain of novel information, creativity (or more specifically the AHA experience) corresponds to the actual gain of novel information. In other words, while in curiosity the model is updated based on future expected gain for the novel information acquired, in creativity the model is updated with a known gain of novel information that was generated, providing the agent a sense of certainty. Future research is yet to determine whether uncertainty reduction and accurate representation of the world is one of the functions that creativity may (or may not, according to Singh & Murayama) serve. However, it is important to consider the prediction error minimization idea when approaching the debate. In accordance with Zhou & Bornstein, creativity may increase certainty and support the pursuit of long-term goals, because once a problem is solved, the uncertainty associated with this problem is reduced. In other words, at least the convergent subtype of creativity involved in problem-solving may serve to reduce uncertainty, similar to the deprivation subtype in curiosity. This link is also supported by Gustafsson et al.'s optimal arousal model according to which both specific curiosity and convergent thinking are aimed at solving problems to simplify one's environment. However, diversive curiosity and divergent thinking are aimed to get more stimulation and to complexify one's environment and thus may not be associated with uncertainty reduction. This debate is likely to unfold by future work that would investigate whether uncertainty reduction is nonetheless driving all forms of creative and curious behaviors.

Another perspective was offered by Grüning & Krueger who propose that creativity moderates whether curious people can engage in information-seeking behavior. This goes beyond our identification of a link between curiosity and creativity and advances to suggest a causal relationship between the two. According to their model, a certain degree of a certain type of curiosity and creativity is prerequisite for exhibiting information-seeking behavior and eventually the two suggested pathways of “information-generating” and “information-gathering” result with the same behavior. This idea is interesting and requires more elaboration and evidence. What are the qualitative differences between these two pathways in terms of outcome? Why does “information-generating” relate to specific curiosity more than diverse curiosity? These open questions are yet to be answered before an unequivocal causal link between creativity and curiosity can be made.

On that note we should emphasize that our proposed NSM does not dismiss other possible variables that might link curiosity and creativity, which seems to have been the concern of several commentators (Litovsky et al.; Holm & Schrater; Runco). We rather provide a novel approach for investigating the proposed link between them. The host of findings we have cited in the target article, together with those provided by the many supportive commentaries, indicate a strong case for why this framework should be pursued and thoroughly tested in the future. We hope that by advancing the theoretical understanding about the link between curiosity and creativity and establishing a testable model, we set the ground for empirical research to elucidate their intuitive yet complex relationship (a challenging mission, as evidenced in Raz & Kennet commentary). Unified definitions and reliable measures of each of the subtypes of curiosity and creativity are necessary to achieve this important goal.