The neuroscience of curiosity
Date: 22-08-2018
Name: Carlo Rooth Student ID: 10198350
Supervisor: Suzanne Oosterwijk Co-assessor: Michiel van Elk
MSc in Brain in Cognitive Sciences, Cognitive Science track, University of Amsterdam
Abstract
Though curiosity is an important motivator in our acquisition of knowledge, it is still an understudied phenomenon. Curiosity is defined as a drive-state that a person wants to reduce. The current literature thesis will investigate what processes are associated with this drive-state. It will be shown that curiosity can be categorized in four different phases that follow each other: salience detection, induction of curiosity, uncertainty reduction and the relief of curiosity. Different neuroimaging studies will be reviewed to explore what the neural correlates of these processes are. It is found that curiosity serves as an intrinsic reward that is associated with activations in the salience network and reward network in the brain. The difference between the predicted value and the actual value of this reward (reward prediction error) enhances memory about curious items. The thesis will end with a discussion on how research on curiosity can be improved.
Table of contents
Intro
A theory of human curiosity - Two dimensions of curiosity
- Curiosity evokes a drive-state that needs to be reduced
- Higher curiosity has positive effect on memory
- Connecting behaviourism and recent disciplines
Different studies on the induction and relief of curiosity
- Epistemic curiosity activates reward circuitry and enhances memory
- Curiosity is monotonically related to uncertainty, and its induction and relief are associated with activations in parietal and insular cortices
- States of curiosity activate reward circuitry and modulate hippocampus-dependent learning
- Relief of perceptual curiosity is associated with activations in reward- and memory related areas
The salience network
- Detecting salient stimuli
- The salience network underlies the first stage in curiosity
Induction of curiosity
- Information serves as an intrinsic reward
- Wanting and liking a reward are related, but distinct processes
Uncertainty reduction
- Uncertainty about a stimulus is essential for curiosity
- The neural correlates of uncertainty
Relief of curiosity
- Reward prediction error enhances learning
- The reward system
- Curiosity activates reward circuitry
- Curiosity enhances reward-learning and activates memory circuitry
Conclusion
Critical discussion on future perspectives
- Different levels of wanting and liking are associated with different forms of curiosity
- Future of paradigms to measure curiosity
The neuroscience of curiosity
Imagine you are in a coffee bar and you are talking with a friend about your holidays. She tells you she went to Tobago. You reply: “Ah, cool! Tobago!” without knowing where Tobago is. You ask yourself the question: “Where in the world is Tobago?”, while the answer remains unknown to you. At this point, curiosity is induced. You find yourself in what is called a drive-state: a state that is produced by a homeostatic disturbance and causes specific behaviour to restore balance (Berlyne, 1954; Seward, 1956). In the theory of Berlyne (1954) curiosity is described as a drive-state that a person wants to reduce. Explorative behaviours and information seeking are used to restore balance. However, this theory does not explain what mental processes are associated with the drive-state. So, now that you are deprived of information, this situation leads us to the question how curiosity arises and what the underlying mechanisms are.
This literature review will unravel the complex topic of curiosity by investigating which neural processes are associated with curiosity. People do not necessarily have awareness on how their behaviours associated with curiosity arise. As such, it is helpful to look at the brain, as it enables us to track processes that subjects are unaware of. In that way, it helps us to understand the underlying mechanisms of curiosity.
First, a behaviouristic theory of human curiosity by Berlyne (1954) will be highlighted. By using recent literature from psychology and neuroscience, we are able to update this classic theory. Then, it will be shown that curiosity is associated with four different phases, and that the drive-state in Berlyne’s theory is associated with salience, reward and memory processes in the brain. The thesis will end with a discussion about the current state of the field and suggestions for future research.
A theory of human curiosity
One of the pioneers in curiosity research is Daniel Berlyne. His paper from 1954 focuses on the question why certain pieces of knowledge are more extensively sought out and remembered than others. The drive behind this process is curiosity (Berlyne, 1954). Berlyne distinguishes curiosity on two dimensions: perceptual vs epistemic curiosity, and specific vs diversive curiosity. According to his theory, curiosity is a drive-state that can be reduced by knowledge-rehearsal.
Berlyne’s definition is based on a distinction between epistemic curiosity and perceptual curiosity on one dimension, and between specific curiosity and diversive curiosity on the other dimension. Berlyne referred to perceptual curiosity as a drive that is aroused by novel stimuli and leads to increased perception of the stimulus by stimulating explorative behaviour in the animal. By exploring the stimulus the animal is continually exposed to it, hence reducing the curiosity for the stimulus (Berlyne, 1954). A similar process is present in humans, and it leads to our concept of knowledge. This can be best explained by a simple example of a young infant. When an infant gets handed a new toy, it stimulates the infant to explore the toy. He wants to touch it, smell it and taste it. Then he has a better perception of the toy. The next time he sees the toy, he won’t explore the toy as extensively as he did the first time. The infant saved a representation of the toy in his mind: the infant has gained ‘knowledge’ about the toy.
Though perceptual curiosity leads to knowledge about stimuli and objects in humans, Berlyne distinguishes it from epistemic curiosity. Epistemic curiosity is the motive to seek, obtain and make use of new knowledge (Berlyne, 1954). This can be explained by the human ability to think and reason on abstract knowledge. For example, when you watch the news and you see images of a battle-field in the Middle-East, it might put questions to you as: “What political issues have caused this war?” Contrary to perceptual curiosity, the question will not be answered if your perception of the stimulus is increased. The drive that is caused by epistemic content can only be reduced by reasoning on other abstract knowledge.
This is what Berlyne calls knowledge-rehearsal. So, epistemic curiosity is a drive-state that is reduced by knowledge-rehearsal. It also strengthens knowledge, because new acquired knowledge works as a reinforcement to reduce uncertainty (Berlyne, 1954). In other words, if a person finds an answer to a question, the answer is a reward for the process of reducing the drive.
The other dimension is specific vs diversive curiosity. If a person is left with lack of information, uncertainty arises with a particular piece of information. The curiosity that is evoked for a particular piece of information is called specific curiosity. On the other hand, when someone seeks out stimulation to find an optimum of arousal, regardless of a certain context, it is called diversive curiosity (Berlyne, 1966). Diversive curiosity is commonly shared under sensation seeking (Berlyne, 1966; Loewenstein, 1994). This is a human trait that is characterized by the need for varied, novel and complex sensations and experiences, and the willingness to take physical and social risks for the sake of such experience (Zuckerman, 1990). Diversive curiosity and sensation seeking seem to reflect a personality trait rather than a state
of curiosity. Personality traits vary a lot across the population and are therefore hard to study. Thus, the rest of the paper will focus only on the specific type of curiosity.
Curiosity evokes a drive-state that needs to be reduced
Curiosity is evoked by what Berlyne calls ‘stimuli’, and ‘motivational stimuli’. The cue-stimuli include the external cue-stimuli in the environment that evoke a question in the first place. For example, when your friend talks about Tobago, this serves a cue-stimulus that puts questions in your mind, such as: “Where in the world is Tobago?” The motivational stimuli reflect a drive state that this question evokes. If you know the answer, the drive to know the answer is directly reduced. If not, the drive will persist. The drive will only arise if the question has some meaning to the person. The more meaning a question has, the higher is the drive, so a person is more curious for the answer (Berlyne, 1954).
These ideas have been further developed by other researchers. The drive in Berlyne’s theory is similar to what Loewenstein calls the gap. According to the information-gap theory, curiosity arises when the attention of an individual is drawn towards a information-gap in one’s knowledge between what one knows and what one wants to know. Such a gap produces a feeling of deprivation that is labelled as curiosity. He notes that curiosity is positively related to one’s knowledge in a particular domain. When some individual gains more knowledge about a topic, it is more likely that this individual is attracted to the gap in one’s knowledge when some information about this topic is missing (Loewenstein, 1994). So, suppose you already know a lot about topography. It is more likely that you will be curious about where Tobago is, because you have the feeling that you miss information about a meaningful topic for you.
The drive or the information-gap is only reduced if the person finds the answer. During this process conflict will arise between an expectancy and a perception. Conflict gives the person an unpleasant feeling of uncertainty. When this feeling is reduced, it is rewarding and pleasurable for the person. Just as it is rewarding to eat when you are hungry, or drink when you are thirsty. The feeling of uncertainty is reduced by gathering information about the relevant stimulus, so that cognitive coherence is restored (Berlyne, 1954; Litman, 2005). So, at this point you will grab your phone or an atlas to look up the location of Tobago.
Higher curiosity has positive effect on memory
When a question has meaning to the person, the drive-state is more intense. The intensity of the drive-state is an important factor on the effects of curiosity. Reinforcement-learning holds that
when a stimulus is perceived, and the subsequent response is rewarded, the response is more likely to occur if the same stimulus is perceived in future occasions.
When an acceptable answer to a question is encountered and has been rehearsed, curiosity is reduced. The higher the drive is before it is reduced, the higher is the amount of reinforcement. The stronger this reinforcement process is, the more likely it is that the same response will occur in the future. In this case the reinforcement process is gathering evidence and the response is recalling the answer. In other words, when curiosity for a question is high, the answer will be remembered better, because the process of reducing the drive was more intense. Curiosity is higher, when the question has more meaning to the person (Berlyne, 1954). So, when a question has more meaning to the person, the answer will be better remembered.
Connecting behaviourism and recent disciplines
The theory of Berlyne provides a framework from which we can explore curiosity. There are two forms of curiosity depending on the content of the stimulus: perceptual and epistemic curiosity. Moreover, curiosity is a drive-state that you want to reduce by gathering information. The more drive a question evokes, the higher is the reinforcement when it is reduced. It follows that stimuli which people are curious about are then better remembered.
Though Berlyne’s views were new for that time, the behavioural approach does not expose what happens between stimuli and response, other than some drive-state that evokes uncertainty which needs to be reduced. With current techniques, we can study what processes are responsible for this drive-state. By reviewing literature from different disciplines, such as psychology and neuroscience, this thesis will investigate what neural processes are associated with curiosity.
Different studies on the induction and relief of curiosity
In curiosity a few phases can be distinguished. Two of these phases are the induction and relief of curiosity. These phases have been investigated by recent studies that will be highlighted in the following section (Gruber, Gelman, & Ranganath, 2014; Jepma, Verdonschot, van Steenbergen, Rombouts, & Nieuwenhuis, 2012; Kang et al., 2009; Lieshout, Vandenbroucke, Mu, & Cools, 2018). This thesis proposes two other phases: salience detection occurs prior to the induction of curiosity and a period of uncertainty reduction follows the induction of curiosity.
First, a stimulus can evoke curiosity if it is detected as a salient event by a person. A stimulus is salient if it is surprising, rewarding, or deviant from other stimuli (Menon, 2015). A
salient event does not necessarily induce curiosity. The salient event needs to evoke an approach intention, and the individual must be uncertain about the outcome of the event. Only then the second phase arises: curiosity induction. The third phase is a period of uncertainty reduction. The individual will further explore this stimulus and extract information from this stimulus to gain knowledge about it, hence reducing the uncertainty. The fourth phase is the relief of curiosity. The individual has found answers to the questions and uncertainty is reduced. This is rewarding to the individual. The next time a similar stimulus evokes these questions, the individual knows the answer right away, since the rewarding process of curiosity enhanced his memory for the questions.
As people are not always able to explicitly report on processes such as reward and the induction of curiosity, it is helpful to look in the mechanism that is assumed to produce these processes: the brain (Kidd & Hayden, 2015). Four different studies have explored the neural correlates of the induction and relief of curiosity. Two studies used a trivia questions paradigm, and one study used a lottery paradigm to evoke epistemic curiosity. One study used a blurred pictures paradigm to evoke perceptual curiosity. Some of these studies also tested the memory of subjects directly after the experiment or 1 – 2 weeks later. In that way, these studies investigated the neural correlates of curiosity and its implications for learning and memory. I will briefly discuss the methods and results of the four papers.
Epistemic curiosity activates reward circuitry and enhances memory
Kang et al. (2009) explored the neural correlates of curiosity and tested hypotheses derived from its findings in two other experiments. In the fMRI scanner subjects were presented with trivia questions that evoked curiosity. Subjects had to read the question, guess the answer in mind, rate their curiosity about the question, and rate the confidence of their guess.
They analysed activations during the presentation of the question, which served as the induction of curiosity and during the appearance of the right answer, which served as the relief of curiosity. In that way, it was analysed what regions were active during the induction and relief of curiosity. The activations were correlated with the curiosity and confidence ratings to analyse the effect of curiosity on neural processes. In a behavioural study 1 – 2 weeks later they tested the memory of the subjects by asking for the answers of questions. These results were correlated with the findings in the scanning session.
First, it was found that in anticipation of the answer the caudate nucleus, prefrontal cortex, and parahippocampal gyri (PHG) were activated. These are regions related to the reward network. This is a network in the brain that is active during processing of rewards and it
underlies learning in humans (Haber & Knutson, 2010; Schultz, 2015). Second, when answers were revealed activations in areas linked to learning and memory were stronger activated if the subject’s guess of the answer was incorrect. These regions included the midbrain, hippocampus and PHG. The strength of the activations was modulated by the level of curiosity, but this was only the case for answers that were guessed incorrect. When subjects guessed correctly, curiosity measures did not correlate with any of the identified regions. Third, higher curiosity was correlated with better recall for answers that were guessed incorrectly in the first session. These results suggest that curiosity is the anticipation of rewarding information and that it enhances memory for learning new and surprising information (Kang et al., 2009).
Curiosity is monotonically related to uncertainty, and its induction and relief are associated with activations in parietal and insular cortices
Lieshout, Vandenbroucke, Mu, & Cools (2018) studied the contribution of uncertainty and the expected value of rewards to curiosity. They designed a lottery task in which a vase with red and blue marbles was shown on a screen. After each trial one marble was drawn from the vase and in 50% of the trials the outcome was shown. Outcome uncertainty was manipulated by the distribution of the marbles, e.g. a vase with 5/20 blue marbles has lower uncertainty than a vase with 10/20 blue marbles. Expected value of a trial was manipulated by the value for each of the marbles, e.g. in some trials blue marbles were awarded with 10 points and in some with 70 points. Curiosity was rated explicitly by rating on a 1 – 4 scale or implicitly by testing the willingness to wait for the answer (3 – 6 seconds). It was hypothesized that people are more curious in high uncertain situations because the information update would be the largest (IPE; information prediction error). The activations during vase presentation and during the lottery presentation were analysed, so that the regions during induction and relief of curiosity could be analysed. These were correlated with the measurements of curiosity. In this study, the effects on memory were not tested.
The results showed more activity in the left inferior parietal lobe with increasing outcome uncertainty at the time of curiosity induction. When curiosity was relieved by showing the outcome of the trial, this resulted in increased activations in the insula, orbitofrontal cortex and parietal cortex. The insula also showed a linear increase to the size of the information update, also referred to the information prediction error. These regions are connected to the reward system and the salience network, which are associated with reward-learning and detecting important stimuli (Lieshout et al., 2018; Menon, 2015).
States of curiosity activate reward circuitry and modulate hippocampus-dependent learning Gruber, Gelman, & Ranganath (2014) investigated how curiosity influences memory. They used a trivia questions paradigm. The experiment consisted of three phases: a screening phase, a scanning phase and a memory test phase. During the screening phase subjects learned the trivia questions and rated how likely they thought it was that they knew the answer and how curious they were for the answer. During the scanning phase subjects were scanned and the questions were shown again, this time with the answer appearing after a 10 second delay. A neutral, unrelated face was shown for 2 seconds in the delay-time. During the memory test phase subjects performed an unexpected memory test for the faces and for the answers to the trivia questions. The memory test was also carried out one day later. Analyses focused on the activation patterns at the presentation of the face and the trivia answer. The activation at the time of answer presentation served as the relief of curiosity. These activations were correlated with the measurements of curiosity on the immediate and one-day-delayed tests.
It was found that memory was improved for information that subjects were curious about. Moreover, activity in the midbrain and the nucleus accumbens was increased when curiosity was higher. These regions are active in the reward network when stimuli are encoded as rewards. At the relief of curiosity activations in the hippocampus were found, suggesting a link between curiosity and memory. Later recall data confirmed this, as subjects recalled the unrelated faces and the answer to the trivia questions better when they were in a high curiosity state. The results suggest there is a link between reward circuitry and curiosity, and that these have positive effects on learning and memory (Gruber et al., 2014).
Relief of perceptual curiosity is associated with activations in reward- and memory related areas
Jepma et al. (2012) investigated what happens in our brain during the induction and relief of perceptual curiosity. Subjects were presented with sequences of two pictures. Perceptual uncertainty was manipulated by using combinations of blurred and clear pictures. If a blurred picture was followed by a corresponding clear picture, this was counted as a trial in which perceptual uncertainty was relieved. Activations were measured during the induction of curiosity at the time of appearance of the first picture, and relief of curiosity at the time of appearance of the second picture. After the scanning session they were asked to name as many objects as they could recall in an unexpected free-recall session.
They found the anterior cingulate cortex (ACC) and insula to be active when the first picture of a sequence was blurred. So, the ACC and insula are associated with the induction of perceptual
curiosity. These regions are also part of the salience network, associated with detecting important stimuli from the environment (Menon, 2015). Moreover, they found strong deactivations in regions related to the default mode network, when the first picture was clear compared to blurred. This suggests subjects actively processed the blurred pictures (Jepma et al., 2012). When curiosity was relieved, the second picture was a corresponding clear picture of a blurred picture. This was associated with activations in the orbitofrontal cortex, left striatum, including the caudate, putamen and nucleus accumbens. These areas are associated with coding of reward values, reward prediction errors and reward learning. Moreover, hippocampal activation showed stronger activation when the second picture was shown in the blurred-clear sequence. This underlies the enhanced later recall of these stimuli. Interestingly, the activation strength in hippocampus at curiosity relief correlated with the activation strength in insula at curiosity induction (Jepma et al., 2012). This suggests that there is a monotonical relation between salience of a stimulus and the memory of this stimulus.
So, these four studies show what happens in the brain during the induction and relief of curiosity. Before curiosity is induced, the salience network is active in detecting salient stimuli in the environment. Moreover, curiosity activates circuitry that are important for processing and prediction of rewards. The corresponding responses are better remembered when people are curious about it. So, curiosity might enhance reward-based learning. All these processes will be discussed in the paper, starting with the first process: salience detection.
The salience network
Berlyne’s theory consists of the idea that curiosity is externally motivated. This means that external stimuli in the environment induce curiosity. When your friend says he was on holiday in Tobago, this is an external stimulus that grabs your attention. That means, your brain detected this stimulus as salient. This part of the thesis explains how stimuli are detected as salient, and how it relates to curiosity.
Detecting salient stimuli
Salience refers to the thing that something in our environment is different and therefore important to focus our attention to. We have two systems regarding salience detection. The first is a fast, automatic mechanism that filters stimuli in the environment based on perceptual features. The second is a higher order mechanism that is used for context-specific stimulus
selection and for focusing the spotlight of attention (Menon, 2015). This seems to be in line with the theory of Berlyne, where a distinction is made between perceptual and epistemic curiosity (Berlyne, 1954). Our primitive system is called upon when we perceive something salient because of its perceptual features. Our curiosity drives us to explore this strange and puzzling thing. For instance, among a collection of patterns, people tend to explore the ones that are perceived as visually complex (Berlyne, 1966). Contrary, the higher-order mechanism of salience detection seems to be in play for epistemic curiosity. For instance, when we stumble upon something that conflicts with our prior knowledge, a prediction-error arises, and it is detected as salient information. As said, the higher-order mechanism is active when we focus our attention. So, this mechanism focuses our attention towards the prediction-error, or in terms of Loewenstein (1994): towards the information-gap. In that way, salience detection is the first in the chain of processes that forms curiosity.
The salience network underlies the first stage in curiosity
These mechanisms are associated with the salience network in the brain. The salience network integrates sensory, emotional and cognitive information to detect the most relevant stimuli, both internally and externally. So, when a car honks or you don’t know the answer to a question, this network is active. The salience network works as an alarm system for the body.
The network is important for the bottom-up detection of salient events by identifying the stimulus that impact the senses in a continuous stream of sensory stimuli. The anterior insula and the anterior cingulate cortex form the core of this network and receive inputs from sensory and limbic systems. The salience network is also important for switching between the default mode network (DMN) and the central executive network (CEN). The DMN is a brain network that is active during rest and involves self-referential mental activity. The CEN is active when we perform tasks and involves cognitively demanding mental activity. It is suggested that the SN plays a role in switching between these networks, to keep attention focused on tasks (Menon, 2015; Menon & Uddin, 2010).
In an early study, an oddball paradigm was used in which subjects were presented with a continuous string of similar stimuli. Every 10, 12 or 14 seconds this string was interrupted with a novel stimulus (the oddball). The researchers found a cortical network to be active that includes the temporal-parietal junction (TPJ), anterior cingulate cortex (ACC) and insula. Within the network, the TPJ plays a role in stimulus evaluation in general, regardless of it being a familiar or a novel stimulus. On the other hand, the ACC and insula were activated exclusively with detecting novel and salient events (Downar, Crawley, Mikulis, & Davis, 2002). This is in
line with later research that also found the ACC and insula to be most important in the salience network (Menon & Uddin, 2010; Seeley et al., 2007).
The salience network is also active when curiosity is induced. Jepma et al. (2012) found that the ACC and anterior insula were active during the processing of the blurred pictures. Moreover, deactivations were found in regions associated with the default mode network (Jepma et al., 2012). These results suggest that the salience network is responsible for the induction of curiosity by detecting the blurred picture as a salient stimulus. The deactivation of the default mode network seems to be in line with the SN’s role of switching between the DMN and CEN. This study did not include an active task, which might explain that no activations were found for regions related to the CEN.
So, it seems that the salience network is active in the first stage of curiosity. Activations in the insula and ACC are associated with the detection of a salient, external stimulus from the environment that is selected for further processing by deactivating the DMN. The next part will research what happens next, after a stimulus is detected as salient.
Induction of curiosity
After a stimulus is detected as salient, it needs to evoke other conditions to induce curiosity. At first, the stimulus needs to have some meaning to the individual, to act as a drive-stimulus. The more a meaning a question has, the more curiosity it evokes. Moreover, the outcome of this stimulus needs to be uncertain. If one would be certain of the outcome, this would not induce curiosity, as the drive to know an answer or to reduce perceptual ambiguity is directly reduced. Finally, the stimulus also needs to evoke an approach intention to explore this stimulus, so that uncertainty is reduced (Berlyne, 1954). So, to induce curiosity a stimulus does not only need to be detected as salient, it also needs to have a meaning to the individual, have an uncertain outcome and it needs to evoke an approach intention to reduce uncertainty. However, since there are not many studies focusing on this approach intention, we will now focus on the first necessary condition for curiosity: the meaning of the stimulus.
Information serves as an intrinsic reward
People are motivated to seek out stimuli that are rewarding to them. This is called reward learning: behaviour that is rewarded, is more likely to be repeated. Behaviour can be intrinsically rewarded and extrinsically rewarded. When extrinsically motivated, people engage in the behaviour to obtain an instrumentally separable consequence, such as food for
animals or money for humans. When people are intrinsically motivated, they engage in the behaviour because they find it inherently satisfying (Di Domenico & Ryan, 2017). That means, the behaviour itself serves as a reward for the individual.
The difference between intrinsic rewards and extrinsic rewards sheds light on the importance of curiosity in reward learning. By using a trivia questions paradigm Murayama & Kuhbandner (2011) studied the effect of monetary rewards on immediate (10 minutes after) and delayed (1 week after) memory performance for interesting and uninteresting questions. The monetary reward served as an extrinsic reward. The answer to the question served as an intrinsic reward with interesting questions categorized as having more intrinsic value than uninteresting questions. So, these questions had more meaning to the subjects.
It was found that performance on the delayed memory test improved when people were given money for the questions. However, this effect disappeared when the interest of the questions was accounted for. People remembered the answers just as much, even though they were not rewarded with money (Murayama & Kuhbandner, 2011). This suggests that extrinsic rewards are ineffective for stimuli that have intrinsic value. This study implicates the importance of curiosity in learning: information for something that is interesting to you, serves as an intrinsic reward.
Wanting and liking a reward are related, but distinct processes
When a question is interesting to you, you want to know the answer. The answer serves as an intrinsic reward that you want and like. However, wanting and liking are two concepts that are highly correlated but separable (Berridge, 1996; Berridge, 2009; Peciña, 2008). Liking refers to the hedonic impact of a reward or how pleasurable a reward is, whereas wanting refers to the motivational incentive value of this reward or how strong a reward evokes motivation (Peciña, 2008).
Pecina, Cagniard, Berridge, Aldridge, & Zhuang (2003) studied the role of dopamine in natural rewards and the distinction between wanting a reward and liking a reward. They manipulated one group of mice, so that these mice have elevated levels of dopamine. They analysed what the consequences of dopamine are on spontaneous food intake, motivation and learning to obtain rewards, and affective liking of a reward. They found that mutated mice had higher levels of food intake. Also, they showed better performance on a task and were more motivated to obtain a reward. However, the mice did not differ in liking levels of the reward, compared to wild mice. These results suggest that dopamine facilitates the wanting of a reward, but not the liking. This means wanting and liking are separable processes.
Moreover, in a review article Peciña (2008) examined the role of opioid neurotransmission in the nucleus accumbens on wanting and liking. The nucleus accumbens is an important region in which opioids exert their rewarding effects. It consists of a core with a shell around it and is part of the ventral striatum. It is suggested that ‘liking’ and ‘wanting’ are anatomically dissociable within the nucleus accumbens. It seems that the rostral-dorsal part of the medial shell is important for both liking and wanting a reward, whereas wanting a reward has more widespread distribution throughout the nucleus accumbens. This suggests that opioid circuits outside of the liking hotspot in the nucleus accumbens may stimulate food intake in mice (Peciña, 2008).
These studies suggest that if you want a reward, it is not necessary to like it. So, you do not need to be interested in Caribbean countries to be curious about where Tobago is. You only want to reduce a feeling of deprivation (Litman, 2005). This distinction will be discussed later in the thesis. As we will see in the next part, after you decide to want a reward, a period of uncertainty arises, before you obtain the reward.
Uncertainty reduction
Curiosity is induced when a stimulus is salient, the outcome is uncertain, and the individual is motivated to seek out the answer, or to reduce ambiguity. What follows is a period in which the individual tries to reduce his uncertainty about the stimulus. The individual will seek out the stimulus or he will wait for the answer to be shown, as is the case in experimental paradigms. In this part it will be shown what processes underly uncertainty reduction.
Uncertainty about a stimulus is essential for curiosity
Uncertainty refers to the state of an organism that lacks information about an event. It is characterized as an aversive state that people are motivated to reduce. This is done by gaining information about the event (Bar-Anan, Wilson, & Gilbert, 2009). When a question is put uncertainty about the answer and question arises. Curiosity is induced if this has meaning to the individual, and he is uncertain about the outcome. If people would be certain about the outcome of a question, this would not induce curiosity, since the drive to know the answer is relieved immediately (Berlyne, 1954). So, a certain amount of uncertainty about an outcome is a necessary condition for curiosity.
However, there does not seem to be an optimal level of uncertainty that induces the most curiosity. Some authors suggest the relationship between uncertainty and curiosity is parabolic.
This means that very high and very low levels of uncertainty induce a low level of curiosity, while a moderate degree of uncertainty induces high levels of curiosity (Berlyne, 1954; Loewenstein, 1994). This is found in a study where confidence about an answer was set to the degree of curiosity (Kang et al., 2009). However, other studies did not find this effect and found outcome uncertainty to have a linear relationship with curiosity. That means the higher the outcome uncertainty was, the higher was curiosity (Lieshout et al., 2018). So, the level of uncertainty has a positive relationship with the level of curiosity. Yet it is unclear what the effect is for high levels of uncertainty.
The neural correlates of uncertainty reduction
To be relieved of curiosity, one has to reduce his uncertainty about the stimulus. The neural processes that underly the reduction of uncertainty have been studied. During tasks in which uncertainty is evoked, multiple studies find frontal and parietal regions to be active. For instance, Critchley, Mathias, & Dolan (2001) measured brain activity during a delay-period between reward-related decisions and their outcome. It was tested how uncertainty modulate this activity. In the experiment play cards were shown with values ranging from 1 – 10. Subjects had to predict whether the next card would be higher or lower than a cue card. The level of uncertainty depended on the probability of the next card being higher or lower. So, if a card with a value of 5 evokes more uncertainty than a card of value 1 or 10. It was found that orbital and medial prefrontal cortex, temporal and parietal cortex were active during this delay-period. The anterior cingulate cortex, the orbital prefrontal cortex and anterior insula modulated the degree of uncertainty. That means that these regions were more active when uncertainty was higher. This is in line with research on salience and reward networks, where the anterior cingulate cortex and anterior insula are associated with detecting salient stimuli (Menon, 2015), and the orbitofrontal cortex codes for the reward value of a stimulus (O’Doherty, 2004).
Other studies also give a prominent role to the insula for processing uncertainty (Lieshout et al., 2018; Singer, Critchley, & Preuschoff, 2009). In a neuroimaging study subjects were performing a similar task as described above by Critchley et al. (2001). It was found that activations in the anterior part of the insula are associated with the computed ‘risk prediction’. This was found when subjects awaited the outcome of a risky decision, for instance when the cue card is 5. Interestingly, the anterior insula was also active at the moment when the outcome was shown, when the second card was presented. At that point a ‘risk prediction error’ is calculated, which is the difference between the predicted risk and the actual risk (Preuschoff, Quartz, & Bossaerts, 2008). This was also found in the study by Lieshout et al. (2018), where
stronger activations in the insula were associated with a larger prediction error. This concept of prediction error will be discussed in the next part and is important when curiosity is relieved after the period of uncertainty reduction.
Relief of curiosity
As the period of uncertainty endures, you are staring in the atlas to find Tobago. At a sudden moment your eye falls on a small island in the Caribbean Sea above Venezuela. You found where Tobago is! Now your curiosity is relieved, as you have found the answer to the question that you have put yourself. The next time, someone asks you where Tobago is, or when you hear about it on the news, you immediately know where it is. Though this answer was not rewarded extrinsically, you do have a pleasant feeling as you enhanced your knowledge: this information served as an intrinsic reward. But how did your brain respond to this process?
Reward prediction error enhances learning
Murayama & Kuhbandner (2011) showed that information serves as an intrinsic reward, if it is interesting to the individual. This had a positive effect on memory, since subjects memorized the answers to interesting questions at the same level as uninteresting questions, though they were only extrinsically rewarded for uninteresting questions. Yet it was unclear whether this effect came from the intrinsic value of the reward itself or that another process underlies the effect. Marvin & Shohamy (2016) investigated this by using a trivia question paradigm. Several trivia questions were shown, and the subjects could choose to skip the answer, wait for the answer, or indicate that they already know the answer. Afterwards, subjects were asked to rate their curiosity for each question and their satisfaction of the answer when it was shown. Using these values, they calculated an information prediction error (IPE): the difference between the actual value of information received (satisfaction) and the anticipated value of the information (curiosity). A positive IPE holds that satisfaction is higher than curiosity, whereas a negative IPE holds the reverse relation. One week after the experiment, subjects were asked to give the answers for the questions to test their memory.
It was found that people were more likely to remember the correct answer when their curiosity for that question was higher and when the IPE was higher. When satisfaction of an answer is greater than the curiosity for the corresponding question, the information is better remembered. Therefore, it is not merely the absolute value of an intrinsic reward that enhances memory, but rather the gap between the expected reward and the actual reward (Marvin &
Shohamy, 2016). This finding is in line with the information gap theory of Loewenstein, in which curiosity is described as a gap in knowledge between what one knows and what one wants to know (Loewenstein, 1994). It is not the piece of information, but rather the drive to close the gap that enhances memory.
The reward system
Marvin & Shohamy (2016) found the information prediction error enhances memory. The information served as a reward, so the more generally used term is reward prediction error (RPE). The RPE has been associated with dopamine signals of neurons in the midbrain. They are more active for positive RPE and are deactivated for negative RPE (Schultz, 2016). The responsible process is the reward pathway in which a part of the midbrain, the ventral tegmental area (VTA), plays a key role. Via the mesolimbic dopamine pathway, the VTA releases dopamine into subcortical regions in the limbic system, such as the nucleus accumbens (NAcc) and the hippocampus. Via the mesocortical pathway the VTA releases dopamine into frontal cortices, such as the orbitofrontal cortex (OFC) (Berridge & Kringelbach, 2015; Haber & Knutson, 2010; O’Doherty, 2004). The NAcc is part of the ventral striatum, an important region in reward expectation and processing reward prediction errors (Daniel & Pollmann, 2014). The OFC is important in encoding reward values (Blanchard, Hayden, & Bromberg-Martin, 2015; O’Doherty, 2004), and the hippocampus is important for forming memories about rewards (Wittmann et al., 2005). Thus, when rewards are processed many regions are activated and connected through the reward system.
Curiosity activates reward circuitry
Curiosity serves as an intrinsic reward prediction. Consistent with this view, parts of the reward system are found to be activated during states of high curiosity. Regions in the midbrain, including the VTA and SN, have been shown to be active when subjects are anticipating a reward (Gruber et al., 2014). This can be explained by the findings that dopamine neurons signal for reward prediction error (Schultz, 2016). Moreover, higher activation in the midbrain and nucleus accumbens were found during states of high curiosity. This suggest curiosity relies on a similar network as extrinsic reward system.
The study by Kang et al. (2009) also found activations in regions related to the reward system. The midbrain, striatum (caudate nucleus and putamen) and prefrontal cortex were more activated when curiosity was higher, and when an answer was incorrectly guessed. This seem to reflect a reward prediction error, because there is a gap between the actual answer and the
anticipated answer (Kang et al., 2009). These results of this study are in line with other studies, in which the reward circuitry are activated for reward prediction errors (Schultz, 2016).
Other studies have found the OFC to be active when curiosity is relieved. When the outcome of the lottery trial was shown, the OFC was active. This might reflect the processing of the reward value (Lieshout et al., 2018). Moreover, in a study with monkeys, it was found that neurons in the OFC encode for both water amount (extrinsic reward) and informativeness of a gamble outcome (intrinsic reward). The OFC neurons fired more, when more value was given to the information, but the OFC did not integrate the values of water and information to code for overall subjective reward value (Blanchard et al., 2015). These studies show the importance of the OFC in reward processing, irrespective whether it is an extrinsic reward or an intrinsic reward.
Concluding, curiosity seems to activate reward circuitry in the brain. This gives rise to the idea that information is processed as an intrinsic reward (Gruber et al., 2014; Kang et al., 2009). These different studies differ in underlying brain mechanisms, and this can be explained by the different paradigms that are used. The last part of the thesis will discuss how future research might take these differences into account.
Curiosity enhances reward-learning and activates memory circuitry
According to Berlyne, when an answer is accepted, curiosity is reduced. But the higher a drive before the reduction, the stronger is the process of reducing this drive. That means that in future occasions the probability of showing this response is higher (Berlyne, 1954). In other words, the more someone is curious about something, the better the answer will be remembered.
As shown earlier in the thesis, this theory still holds. Trivia questions that were interesting for the subjects were better remembered than uninteresting questions (Murayama & Kuhbandner, 2011). Moreover, it seems that the reward prediction error enhances this learning effect (Marvin & Shohamy, 2016). The discussed neuroimaging studies give neural evidence for this. For instance, the study of Kang et al. (2009) found that subjects could recall answers better when they were curious about it. The level of curiosity modulated activations in regions that are important for memory. The hippocampus was more active when subjects guessed answers incorrectly, than when they guessed an answer correctly. These activations correlated with the behavioural measures of recall (Kang et al., 2009).
Gruber et al. (2014) found similar effects. Recall measures for both answers to the trivia questions, as well as for the unrelated faces were better when subjects were in a high state of curiosity. These learning effects were related to more brain activity in the mesolimbic
dopaminergic circuit when subjects were more curious about the answer. These regions included the VTA, nucleus accumbens and hippocampus (Gruber et al., 2014).
Also Jepma et al. (2012) found these effects for perceptual curiosity. When the blurred picture was followed by a clear picture, hence relieving perceptual curiosity, stronger activations in the hippocampus were found when the clear picture was shown. This activation underlies the enhanced recall of these pictures (Jepma et al., 2012). So, this enhanced memory-effect for curious items seems to be the result of the reward network, as the hippocampus is active in the relief of both epistemic curiosity (Gruber et al., 2014; Kang et al., 2009) and perceptual curiosity (Jepma et al., 2012), while other reward circuitry are also activated, such as the striatum and the midbrain.
Berlyne already stated in 1954 that a stronger reinforcement process enhances memory, and that curiosity makes this process stronger (Berlyne, 1954). With current techniques and recent literature, we can see that this reinforcement process is driven by the reward network. The VTA releases dopamine into the striatum and frontal cortices, which fire back to the VTA and hippocampus. Neuroimaging studies on curiosity also find this reward circuitry to be active during the relief of curiosity. It is suggested that these patterns underlie the learning effect for curious items, and that this effect is stronger depending on the size of the reward prediction error. So, these results suggest there is a link between the mechanisms supporting extrinsic rewards and intrinsic curiosity, which has benefits on learning and memory. It is mainly the reward prediction error that drives this effect.
Conclusion
Back to the coffee bar where you are talking to your friend. As she was talking about her holiday in Tobago, you were curious where Tobago is. You came in a drive-state of curiosity that you wanted to reduce. In the previous part it was shown what the processes were that formed your curiosity.
At first, you detected something in the conversation that shifted your attention to something. Your brain detected a stimulus as salient, because it was surprising, rewarding or deviant in a constant stream of other stimuli. This was the case because you never heard of a country that is called Tobago before. At that moment the salience network in your brain was active. The insula and anterior cingulate cortex were active and suppressed the default mode
network in your brain (Menon, 2015). So, at that point you could focus your attention on exploring where Tobago is.
Then, you encoded the stimulus as being rewarding. You really want to know where Tobago is. That means the question will motivate you to look for the answer, because it serves as an intrinsic reward. When you wanted this reward, it stimulated dopamine release throughout your brain, but the fact that you like to know this information is only represented in a few hedonic hotspots, such as the rostral-dorsal part of the nucleus accumbens (Peciña, 2008).
The outcome of the question was uncertain, and you wanted to reduce this uncertainty: you want to know where it is, you also like this information, but you don’t know it yet. The uncertainty you have about the stimulus is essential to be curious about it because it will motivate you to gather information to reduce this uncertainty and find the answer. This was probably associated with activations in orbital prefrontal cortex and insula, as you are coding the reward value of the question (Critchley et al., 2001). The salience network also plays a role during uncertainty. When you have found small pieces of information related to the question, these are detected salient. For example, when you thought Tobago sounded like an African country, you might have checked the map of Africa to find Tobago is not there. This piece of information accumulated to the answer; you still don’t know where it is, but you have ruled out that it is an African country.
After staring in the atlas, you found the answer: Tobago is a small island in the Caribbean Sea. At this time your curiosity is relieved. It might have been a surprise that Tobago is not in Africa, but in the Caribbean. This reflects a reward prediction error, and it enhances your memory when the prediction error is larger (Marvin & Shohamy, 2016). In your brain reward circuitry are active. The ventral tegmental area in the midbrain releases dopamine to the striatum and orbitofrontal cortex, which in turn project back to the hippocampus (Haber & Knutson, 2010). The stronger this process is, the better you will remember the answer. This is the case when the subject has more meaning to you and when the prediction error is larger. For instance, when you are planning to go on a holiday, this information has more meaning to you. Or when you always thought Tobago was an African country, you are very surprised to find out it’s not. Then this process is stronger, and you will remember the answer better.
The next time you are overhearing a conversation about Tobago, the described process will be active from the start. The salience network in your brain will be active, but now it will not induce curiosity, because you will immediately know where Tobago is.
The thesis so far has been described what neural processes are underlying curiosity. Curiosity was defined as a drive-state that one wants to reduce, and it was shown that curiosity is associated with reward processes in the brain. The following part will briefly discuss an alternative explanation of curiosity, shortcomings in the current literature and implications for future research.
Different levels of wanting and liking are associated with different forms of curiosity
The thesis showed that during curiosity a piece of information is encoded as a reward to close a gap between what one knows and what one wants to know. The individual wants to have this information and explores it in order to close the knowledge gap. When the knowledge gap is closed, it has a rewarding and pleasant feeling (Loewenstein, 1994). This implicates that when someone wants a reward, he also likes it when this reward is obtained.
However, it was also shown that wanting and liking are two concepts that are distinct (Berridge, 1996; Peciña, 2008; Pecina et al., 2003). Therefore, it is suggested that different levels of wanting and liking are associated with different types of curiosity. In the I/D model by Litman (2005) a distinction is made between two types of curiosity: curiosity as a feeling of deprivation (CFD) and curiosity as a feeling of interest (CFI). ‘Curiosity as a feeling of deprivation’ (CFD) is aroused when individuals feel as if they are deprived of information, and wish to reduce or eliminate their ignorance (Litman, 2005, p. 799). It is similar to definitions of curiosity by Loewenstein (1994), in which curiosity is viewed as a gap in knowledge that you want to close. It is proposed that some negative affections related to uncertainty are involved with CFD, such as frustration or dissatisfaction. Also, CFD evokes more explorative behaviours, as the individual needs closure.
The other type of curiosity in the I/D model is ‘curiosity as a feeling of interest’. It is aroused when individuals do not feel particularly deficient of information, but would nevertheless like to learn something new (Litman, 2005, p. 799). Contrary to CFD, it is proposed that CFI is involved with positive feelings of interest and joy brought by the anticipation of learning something new. Contrary to CFD, CFI does not evoke high levels of exploration, as it is associated with an ‘take-it or leave-it’ approach. That means it does not evoke negative feelings when the new information is not learned.
Therefore, these types of curiosity are associated with different levels of wanting and liking. CFD is associated with a high level of wanting and a high level of liking, because the absence of relevant information stimulates an intense desire for this particular piece of
information. On the other hand, CFI is associated with a low level of wanting and a high level of liking. It is not motivated by a particular piece of knowledge, so there is no gap in knowledge that is associated with a drive-state that you want to reduce. Therefore, the level of wanting is lower (Litman, 2005).
Moreover, it is suggested that a low level of liking and a high level of wanting evokes a state of morbid curiosity. It motivates people to seek information that may be disliked or evoke unpleasant experiences (Litman, 2005). This is shown in two behavioural studies. The first consisted of a choice-paradigm in which subjects had to choose between a neutral/positive picture or a negative picture based on briefly shown visual cues or verbal cues. It was found that people preferred to see the negative pictures over the neutral pictures (Oosterwijk, 2017). Another study found that people even choose to expose themselves to electric shocks. Subjects were presented with a board of electric-shock pens of which some had a certain outcome of giving an electric shock, some had a certain outcome of not giving an electric shock, and some had an uncertain outcome. It was found that subjects preferred to click the pens with an uncertain outcome over the pens with a certain outcome (Hsee & Ruan, 2016). These studies suggest that people want to relieve their curiosity, even if there are negative consequences involved.
Thus, wanting and liking a reward are distinct processes that are associated with different forms of curiosity. A phenomenon such as morbid curiosity has potential to further explore the distinction between wanting and liking.
Future of paradigms to measure curiosity
The theory of Berlyne makes a clear distinction between perceptual and epistemic curiosity: perceptual curiosity is referred to as a drive that is aroused by novel stimuli and leads to increased perception of the stimulus, whereas epistemic curiosity is referred to the motive to seek, obtain and make use of new knowledge (Berlyne, 1954). Perceptual curiosity can be explained by being curious about the perceptual features of a stimulus, whereas epistemic curiosity can be explained as the human ability to be curious about abstract knowledge.
Though many researchers use this distinction, the line between perceptual and epistemic curiosity is not as sharp as it is often drawn. Perceptual features always have a connection with epistemic knowledge. For instance, classifying fruits in different categories is an example of abstract knowledge. When we might be curious whether an orange is classified as a citrus fruit or a forest fruit, we do not only think of its epistemic features, but we also compare perceptual features. Based on taste and appearance, we can think an orange has more in common to other
citrus fruits such as a lemon or grapefruit, so that we can conclude an orange is most likely a citrus fruit. Thus, when curiosity is induced, it is not always clear whether it should be classified as perceptual or as epistemic.
If perceptual and epistemic curiosity would be two distinct concepts, it might also be reflected in the brain by different neural correlates. To date only a few neuroimaging studies investigating curiosity have been carried out, and this thesis has discussed most of them. Moreover, these studies have only focused on only one of the two forms of curiosity. This means that the differences between perceptual and epistemic curiosity are not directly measured. A design in which both perceptual and epistemic curiosity are studied might be helpful to eliminate this shortcoming. By making minimal changes in one task, researchers can manipulate perceptual and epistemic curiosity. Activations in a perceptual condition can then be directly contrasted with activations in an epistemic condition. This might result in a better understanding in the difference between perceptual and epistemic curiosity.
Besides differences in stimuli that induce curiosity, there might be differences in the strategies to relieve uncertainty that is evoked. Huettel (2005) found that the neural processes underlying uncertainty depend on the type of paradigms. In paradigms where learned associations between a stimulus and response are central, they found medial frontal lobes to be more active with increasing uncertainty.
It differs from accumulative paradigms, in which evidence is accumulated to gather evidence for a decision. Uncertainty in these paradigms rely more on dorsolateral prefrontal cortex, posterior parietal cortex and insular regions (Huettel, 2005). These are regions that are important in executive processes, such as working memory (Pessoa, Gutierrez, Bandettini, & Ungerleider, 2002). It is likely that such an accumulative paradigm requires more executive processes, causing the different activation patterns (Huettel, 2005).
It might depend on the type of curiosity whether the accumulative paradigm or the learned associations paradigm is more applicable. If the stimulus and the raised question is of an epistemic nature, the accumulative paradigm might be more applicable. This is the case, because when a question is put, and one is curious for the answer, the person will try to reduce uncertainty by looking for information and gather more and more evidence for a certain answer. Moreover, such a paradigm also includes an approach motive to seek for the answer. These shortcomings are not taken care of in current studies. Most studies use a trivia-question paradigm in which the curiosity of subjects is relieved by a one-time appearance of the answer, and in which an approach motive is not needed to relief curiosity. This might not explain the
process of epistemic curiosity as it is normally experienced outside the controlled settings in the lab.
So, future studies should take care of these addressed shortcomings. This can be done by using paradigms that can directly test differences between perceptual and epistemic curiosity and using accumulative paradigms for testing curiosity of epistemic nature.
Behaviours during uncertainty reduction
In the current thesis, curiosity is separated in four phases: salience detection of a stimulus, curiosity induction, uncertainty reduction and curiosity relief. Moreover, it is noted that behavioural intention to approach or explore the stimulus is needed for curiosity to arise. These behaviours might take place in the third phase, when a person is trying to reduce his uncertainty.
However, this process might be more detailed than described. An individual needs to rely on both automatic processes, such as approach intentions, and on controlled processes, such as cognitive flexibility to reduce uncertainty. The function of approach behaviour is to move an animal or person towards a goal that is rewarding (Corr, 2013). Cognitive flexibility might be described as the ability to change perspectives (Diamond, 2013). These behaviours might be helpful, especially in the context of epistemic curiosity. For example, when one is curious about the beginnings of the Second World War, he first needs to approach an encyclopaedia or history book. Then when he is gathering information, he needs to change his perspective to the zeitgeist of that period to understand how such a war originated. Future research should take this behaviour intention into account when investigating curiosity.
The goal of this thesis was to investigate what neural processes are underlying the drive-state that is associated with curiosity. This was done by separating curiosity in four phases. It was shown that these phases are associated with salience, reward, and memory processes in the brain. To have a better understanding on the neuroscience of curiosity, it might be helpful for future research to look more detailed into the underlying processes within each of the phases. Because as Albert Einstein said: “ Never stop questioning … Never lose a holy curiosity.” (Einstein, 1955, p. 64).
References
Bar-Anan, Y., Wilson, T. D., & Gilbert, D. T. (2009). The Feeling of Uncertainty Intensifies Affective Reactions. Emotion, 9(1), 123–127. https://doi.org/10.1037/a0014607
Berlyne, D. E. (1954). A theory of human curiosity. Journal of Psychology, 45(3), 180–191. Berlyne, D. E. (1966). Curiosity and exploration. Science, 153(3731), 25–33.
Berridge, K. C. (1996). Food reward:Brain substrates of wanting and liking. Neuroscience Biobehav. Rev., 20(1), 1–25. https://doi.org/10.1016/0149-7634(95)00033-B
Berridge, K. C. (2009). Wanting and liking: Observations from the neuroscience and psychology laboratory. Inquiry, 52(4), 378–398.
https://doi.org/10.1080/00201740903087359
Berridge, K. C., & Kringelbach, M. L. (2015). Pleasure systems in the brain. Neuron, 86(3), 646–64. https://doi.org/10.1016/j.neuron.2015.02.018
Blanchard, T. C., Hayden, B. Y., & Bromberg-Martin, E. S. (2015). Orbitofrontal cortex uses distinct codes for different choice attributes in decisions motivated by curiosity. Neuron, 85(3), 602–614. https://doi.org/10.1016/j.neuron.2014.12.050
Corr, P. J. (2013). Approach and avoidance behaviour: Multiple systems and their interactions. Emotion Review, 5(3), 285–290.
https://doi.org/10.1177/1754073913477507
Critchley, H. D., Mathias, C. J., & Dolan, R. J. (2001). Neural activity in the human brain relating to uncertainty and arousal during anticipation. Neuron, 29(2), 537–545. https://doi.org/10.1016/S0896-6273(01)00225-2
Daniel, R., & Pollmann, S. (2014). A universal role of the ventral striatum in reward-based learning: Evidence from human studies. Neurobiology of Learning and Memory, 114, 90–100. https://doi.org/10.1016/j.nlm.2014.05.002
Di Domenico, S. I., & Ryan, R. M. (2017). The Emerging Neuroscience of Intrinsic Motivation: A New Frontier in Self-Determination Research. Frontiers in Human Neuroscience, 11(March), 1–14. https://doi.org/10.3389/fnhum.2017.00145
Diamond, A. (2013). Executive Functions. Annual Review of Psychology, 64(1), 135–168. https://doi.org/10.1146/annurev-psych-113011-143750
Downar, J., Crawley, A. P., Mikulis, D. J., & Davis, K. D. (2002). A Cortical Network Sensitive to Stimulus Salience in a Neutral Behavioral Context Across Multiple Sensory Modalities. Journal of Neurophysiology, 87(1), 615–620.
https://doi.org/10.1152/jn.00636.2001
books.google.com
Gruber, M. J., Gelman, B. D., & Ranganath, C. (2014). States of Curiosity Modulate Hippocampus-Dependent Learning via the Dopaminergic Circuit. Neuron, 84(2), 486– 496. https://doi.org/10.1016/j.neuron.2014.08.060
Haber, S. N., & Knutson, B. (2010). The reward circuit: Linking primate anatomy and human imaging. Neuropsychopharmacology, 35(1), 4–26. https://doi.org/10.1038/npp.2009.129 Hsee, C. K., & Ruan, B. (2016). The Pandora Effect: The Power and Peril of Curiosity.
Psychological Science, 27(5), 659–666. https://doi.org/10.1177/0956797616631733 Huettel, S. A. (2005). Decisions under Uncertainty: Probabilistic Context Influences
Activation of Prefrontal and Parietal Cortices. Journal of Neuroscience, 25(13), 3304– 3311. https://doi.org/10.1523/JNEUROSCI.5070-04.2005
Jepma, M., Verdonschot, R. G., van Steenbergen, H., Rombouts, S. A. R. B., & Nieuwenhuis, S. (2012). Neural mechanisms underlying the induction and relief of perceptual curiosity. Frontiers in Behavioral Neuroscience, 6(February), 1–9.
https://doi.org/10.3389/fnbeh.2012.00005
Kang, M. J., Hsu, M., Krajbich, I. M., Loewenstein, G., McClure, S. M., Wang, J. T., & Camerer, C. F. (2009). The wick in the candle of leraning. Psychological Science, 20(8), 963–974. https://doi.org/10.1111/j.1467-9280.2009.02402.x
Kidd, C., & Hayden, B. Y. (2015). The Psychology and Neuroscience of Curiosity. Neuron, 88(3), 449–460. https://doi.org/10.1016/j.neuron.2015.09.010
Lieshout, L. L. F. Van, Vandenbroucke, A. R. E., Mu, N. C. J., & Cools, R. (2018). Induction and Relief of Curiosity Elicit Parietal and Frontal Activity. Journal of Neuroscience, 38(10), 2579–2588. https://doi.org/10.1523/JNEUROSCI.2816-17.2018
Litman, J. A. (2005). Curiosity and the pleasures of learning: Wanting and liking new information. Cognition and Emotion, 19(6), 793–814.
https://doi.org/10.1080/02699930541000101
Loewenstein, G. (1994). The psychology of curiosity: A review and reinterpretation. Psychological Bulletin. https://doi.org/10.1037/0033-2909.116.1.75
Marvin, C. B., & Shohamy, D. (2016). Curiosity and reward: Valence predicts choice and information prediction errors enhance learning. Journal of Experimental Psychology: General, 145(3), 266–272. https://doi.org/10.1037/xge0000140
Menon, V. (2015). Salience Network. Brain Mapping: An Encyclopedic Reference, 2, 597– 611. https://doi.org/10.1016/B978-0-12-397025-1.00052-X
model of insula function. Brain Structure & Function, 214(5–6), 655–667. https://doi.org/10.1007/s00429-010-0262-0
Murayama, K., & Kuhbandner, C. (2011). Money enhances memory consolidation - But only for boring material. Cognition, 119(1), 120–124.
https://doi.org/10.1016/j.cognition.2011.01.001
O’Doherty, J. P. (2004). Reward representations and reward-related learning in the human brain: Insights from neuroimaging. Current Opinion in Neurobiology, 14(6), 769–776. https://doi.org/10.1016/j.conb.2004.10.016
Oosterwijk, S. (2017). Choosing the negative: A behavioral demonstration of morbid curiosity. PLoS ONE, 12(7), 1–20. https://doi.org/10.1371/journal.pone.0178399 Peciña, S. (2008). Opioid reward “liking” and “wanting” in the nucleus accumbens.
Physiology and Behavior, 94(5), 675–680. https://doi.org/10.1016/j.physbeh.2008.04.006
Pecina, S., Cagniard, B., Berridge, K. C., Aldridge, J. W., & Zhuang, X. (2003).
Hyperdopaminergic Mutant Mice Have Higher “ Wanting ” But Not “ Liking ” for Sweet Rewards. The Journal of Neuroscience, 23(28), 9395–9402.
https://doi.org/10.1097/00008877-200409000-00094
Pessoa, L., Gutierrez, E., Bandettini, P., & Ungerleider, L. (2002). Neural correlates of visual working memory: fMRI amplitude predicts task performance. Neuron, 35(5), 975–987. https://doi.org/10.1016/S0896-6273(02)00817-6
Preuschoff, K., Quartz, S. R., & Bossaerts, P. (2008). Human Insula Activation Reflects Risk Prediction Errors As Well As Risk. Journal of Neuroscience, 28(11), 2745–2752. https://doi.org/10.1523/JNEUROSCI.4286-07.2008
Schultz, W. (2015). Neuronal Reward and Decision Signals: From Theories to Data. Physiological Reviews, 95(3), 853–951. https://doi.org/10.1152/physrev.00023.2014 Schultz, W. (2016). Dopamine reward prediction error coding. Dialogues in Clinical
Neuroscience, 18(1), 23–32. https://doi.org/10.1111/clr.41_12679
Seeley, W. W., Menon, V., Schatzberg, A. F., Keller, J., Glover, G. H., Kenna, H., … Greicius, M. D. (2007). Dissociable Intrinsic Connectivity Networks for Salience Processing and Executive Control. Journal of Neuroscience, 27(9), 2349–2356. https://doi.org/10.1523/JNEUROSCI.5587-06.2007
Seward, J. P. (1956). Drive, incentive, and reinforcement. Psychological Review, 63(3), 195– 203. https://doi.org/10.1037/h0048229
empathy and uncertainty. Trends in Cognitive Sciences, 13(8), 334–340. https://doi.org/10.1016/j.tics.2009.05.001
Wittmann, B. C., Schott, B. H., Guderian, S., Frey, J. U., Heinze, H. J., & Düzel, E. (2005). Reward-related fMRI activation of dopaminergic midbrain is associated with enhanced hippocampus-dependent long-term memory formation. Neuron, 45(3), 459–467. https://doi.org/10.1016/j.neuron.2005.01.010
Zuckerman, M. (1990). The Psychophysiology of Sensation Seeking. Journal of Personality, 58(1), 313–345.
