ArticlePDF Available

A transdisciplinary view on curiosity beyond linguistic humans: animals, infants, and artificial intelligence

January 2024
Biological reviews of the Cambridge Philosophical Society 99(3)

DOI:10.1111/brv.13054

License
CC BY 4.0

Authors:

Sofia I F Forss

University of Zurich

Alejandra Ciria

National Autonomous University of Mexico

Show all 6 authorsHide

Curiosity is a core driver for lifelong learning, problem-solving and decision-making. In a broad sense, curiosity is defined as the intrinsically motivated acquisition of novel information. Despite a decades-long history of curiosity research and the earliest human theories arising from studies of laboratory rodents, curiosity has mainly been considered in two camps: 'linguistic human' and 'other'. This is despite psychology being heritable, and there are many continuities in cognitive capacities across the animal kingdom. Boundary-pushing cross-disciplinary debates on curiosity are lacking, and the relative exclusion of pre-linguistic infants and non-human animals has led to a scientific impasse which more broadly impedes the development of artificially intelligent systems modelled on curiosity in natural agents. In this review, we synthesize literature across multiple disciplines that have studied curiosity in non-verbal systems. By highlighting how similar findings have been produced across the separate disciplines of animal behaviour, developmental psychology, neuroscience, and computational cognition, we discuss how this can be used to advance our understanding of curiosity. We propose, for the first time, how features of curiosity could be quantified and therefore studied more operationally across systems: across different species, developmental stages, and natural or artificial agents.

Glossary of terms potentially interlinked with curiosity, or the ways curiosity is measured.

…

Figures - uploaded by Sofia I F Forss

Content may be subject to copyright.

Content uploaded by Sofia I F Forss

Content may be subject to copyright.

A transdisciplinary view on curiosity beyond

linguistic humans: animals, infants, and

artiﬁcial intelligence

Soﬁa Forss

1,2,

*,Alejandra Ciria

,Fay Clark

,Cristina-loana Galusca

,David Harrison

and Saein Lee

7,8

Collegium Helveticum, Institute for Advanced Studies, University of Zurich, ETH Zurich and Zurich University of the Arts, Zurich, Switzerland

Department of Evolutionary Biology and Environmental Studies, University of Zurich, Zurich, Switzerland

School of Psychology, Universidad Nacional Autonoma de México, Mexico City, Mexico

School of Psychological Science, University of Bristol, Bristol, UK

Laboratoire de Psychologie et NeuroCognition, CNRS Université Grenoble Alpes, Grenoble, France

Department of History and Philosophy of Science, University of Cambridge, Cambridge, UK

Interdisciplinary Program of EcoCreative, Ewha Womans University, Seoul, Republic of Korea

Department of Psychology, University of Zurich, Zurich, Switzerland

ABSTRACT

Curiosity is a core driver for life-long learning, problem-solving and decision-making. In a broad sense, curiosity is

deﬁned as the intrinsically motivated acquisition of novel information. Despite a decades-long history of curiosity

research and the earliest human theories arising from studies of laboratory rodents, curiosity has mainly been considered

in two camps: ‘linguistic human’and ‘other’. This is despite psychology being heritable, and there are many continuities

in cognitive capacities across the animal kingdom. Boundary-pushing cross-disciplinary debates on curiosity are lacking,

and the relative exclusion of pre-linguistic infants and non-human animals has led to a scientiﬁc impasse which more

broadly impedes the development of artiﬁcially intelligent systems modelled on curiosity in natural agents. In this review,

we synthesize literature across multiple disciplines that have studied curiosity in non-verbal systems. By highlighting how

similar ﬁndings have been produced across the separate disciplines of animal behaviour, developmental psychology,

neuroscience, and computational cognition, we discuss how this can be used to advance our understanding of curiosity.

We propose, for the ﬁrst time, how features of curiosity could be quantiﬁed and therefore studied more operationally

across systems: across different species, developmental stages, and natural or artiﬁcial agents.

Key words: curiosity, intrinsic motivation, exploration, information processing, developmental psychology, animal cogni-

tion, computational cognition.

CONTENTS

I. Introduction .........................................................................2

II. How the history of deﬁnitions has impeded studies on non-human curiosity .........................3

III. Curiosity beyond linguistic humans ........................................................4

(1) Curiosity in animals ............................................................... 4

(a) Novel object/stimulus paradigm .................................................. 5

(b) Problem-solving paradigm ....................................................... 5

(d) Animal curiosity, ecology, and habitat .............................................. 6

*Author for correspondence (Tel.: +41787490158; E-mail: soﬁa.forss@ieu.uzh.ch).

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,

provided the original work is properly cited.

Biol. Rev. (2024), pp. 000–000. 1

doi: 10.1111/brv.13054

(e) Neuroscientiﬁc research on animals ................................................ 7

(2) Curiosity in pre-verbal infants ....................................................... 7

(a) Curiosity from a developmental perspective ......................................... 7

(b) Neuroscientiﬁc research on infants ................................................ 10

(3) Computational modelling of curiosity in artiﬁcial systems ................................. 10

(a)Deﬁnitions and typologies in computational modelling of curiosity ....................... 11

(b) Prediction error: the gold standard measure ........................................ 11

(i) Non-embodied machines ................................................... 12

(ii) Embodied machines ...................................................... 13

IV. Discussion ..........................................................................13

(1) Disciplinary overlaps and their implication for current views on curiosity ..................... 13

(2) Do we need a dichotomic view of curiosity? ........................................... 15

(3) Transdisciplinary future approaches ................................................. 15

V. Conclusions .........................................................................16

VI. Acknowledgements ...................................................................16

VII. Author contributions ..................................................................16

VIII. References ..........................................................................16

I. INTRODUCTION

Curiosity is an intrinsic drive to acquire novel information

and is seen as a deﬁning human characteristic (Barto, 2013;

Markey & Loewenstein, 2014; Dean, Tulsiani & Gupta, 2020).

Curiosity and learning are fundamentally linked; learning

is deﬁned as a relatively permanent change in behaviour due

to lived experience (Washburne, 1936;Lachman,1997), so

curiosity drives the experiences from which we need to learn.

Furthermore, curiosity is fundamentally linked to exploration

and can be seen as the driving force behind it (Berlyne, 1950).

Curiosity is often associated with ‘humanness’.Fromwatching

YouTube videos, reading books, or travelling, many of us

regularly engage in activities primarily fuelled by curiosity

that do not have an extrinsic reward or immediate rele-

vance to our survival (Markey & Loewenstein, 2014). This

constant drive to explore and make sense of our surround-

ings is crucial for educational and creative activities, but

also is seen evolutionarily as a capacity that enabled

humans to be highly adaptive and survive in challenging

environments (Gallagher & Lopez, 2007; Hardy III,

Ness & Mecca, 2017;Shin&Kim,2019). Despite a human

focus, curiosity is in fact not unique to our species, and to

understand better the evolutionary signiﬁcance of curios-

ity and its practical application to artiﬁcially intelligent

systems, we must give more credence to curiosity in all its

forms.

Most empirical assessments of curiosity to date have been

on humans performing language-based tasks. Pre-linguistic

infants/children and non-human animals (hereafter animals)

have been overlooked, which is ironic given that the earliest

studies of curiosity were performed on laboratory rodents

(Berlyne, 1955,1966; Glickman & Hartz, 1964). Perhaps

because older humans have the capacity to self-report their

subjective experience of curiosity (i.e. describe thoughts and

feelings), behavioural observations of curiosity have been rel-

atively overlooked. These taxonomic and methodological

restrictions have led to a failure to evaluate the potential

continuity of curiosity across phylogeny. But doing so will

be key to our understanding of evolutionary selection pres-

sures shaping it. Moreover, focusing on linguistic humans

also overlooks the role of curiosity in infants and younger

children, even though the pressure to acquire novel informa-

tion and adapt to the environment is considerably higher

early in life (Twomey & Westermann, 2018; Oudeyer &

Smith, 2016). Our current restricted knowledge on curiosity

as a mental tool, critical for learning in developing individ-

uals, is striking and research taking the developmental

perspective of such traits will uncover its adaptive function

in humans and other animals.

The overarching aim of this review is to put forward the

scientiﬁc study of curiosity, across disciplines that tradition-

ally have been siloed by differential methods and standpoints.

We reveal how the status quo in curiosity research (i.e. a focus

on linguistic humans) has led to signiﬁcant gaps in knowl-

edge. To do so, we dedicate the ﬁrst part of the article to

showing that the terminology and deﬁnitions surrounding

curiosity partly have philosophical and/or historical reasons,

which divided the topic into two camps with little intercon-

nection between research ﬁelds. We then discuss in what

ways curiosity is conﬂated with other terms, i.e. intrinsic

motivation and exploration tendency, and provide a sum-

mary of the various deﬁnitions of curiosity featured in the

disciplinary literature. Our main contribution is to describe

what is currently known about curiosity beyond linguistic

humans, and how a greater focus on diverse systems will

advance the ﬁeld by paying attention to the biological

functions of curiosity in non-linguistic beings. Ultimately,

we argue that curiosity is not a single trait, but has multiple

components, comparable to a modern view of intelligence

(Gottfredson, 1998; Nyborg, 2003; Warner, 2007). Finally,

we propose future directions for studying curiosity within

a comparative framework, moving beyond traditional

disciplinary borders.

2Soﬁa Forss and others

1469185x, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/brv.13054 by Universitätsbibliothek Zuerich, Wiley Online Library on [30/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

II. HOW THE HISTORY OF DEFINITIONS HAS

IMPEDED STUDIES ON NON-HUMAN

CURIOSITY

It is difﬁcult to parse the relative contributions of animal and

human empirical research in the earliest works on curiosity,

where initial studies may have been undertaken on rodents

then theories subsequently contextualized for humans. We

do not provide a very detailed historical account here but

instead highlight the most salient points. It is difﬁcult to

pinpoint exactly when the scientiﬁc study of curiosity began,

but in the 19th century, William James (commonly referred

to as the ‘father of psychology’) described curiosity as the

motivation to learn what one does not know (James, 1890) and

set the stage in ﬁling curiosity into two distinct types: stand-

point 1 acknowledged a biological function of curiosity as

an instinct-driven behaviour, whereas standpoint 2 was more

complex and posited that curiosity relies on language. By the

beginning of the 20th century, curiosity was widely seen as a

basic biological drive in humans and animals alike, alongside

other drives like hunger or sexual appetite. Curiosity as a

drive to explore was based on animal studies documenting

their exploratory behaviours in the absence of any external

rewards (usually food, e.g. Nissen, 1931).

A paradigm shift came when Berlyne (1960) moved away

from curiosity as a basic biological drive, but like James also

distinguished between two types of curiosity. First, Berlyne

distinguished perceptual from epistemic curiosity. Perceptual

curiosity refers to interest in novel, strange or ambiguous per-

ceptual stimuli which motivates visual and sensory inspec-

tion. By contrast, epistemic curiosity is the desire for

knowledge or intellectual information (which applied mainly

to humans). Later, Berlyne developed these concepts into

diversive and speciﬁc curiosity. For Berlyne, diversive curios-

ity was an individual’s‘taste for adventure’, despite the risks

this may entail. On the other hand, speciﬁc curiosity is the

motivation to explore a speciﬁc object or problem with

the goal of understanding it, thus resembling his initial view

of perceptual curiosity. Berlyne also introduced the concept

of various ‘collative variables’that can trigger human curios-

ity, such as novelty, uncertainty, or complexity, and as we will

see, today these factors are integrated into research on curios-

ity across model systems. Berlyne’s theory proposed that curi-

osity requires an individual to use their cognition to compare

different sources of information: newly perceived informa-

tion against stored memories and understanding. Several

other prominent curiosity theories arose from Berlyne’s

theoretical framework. For example, Loewenstein (1994)

suggested that an information gap was the fundamental

pre-requisite of curiosity; there is a disparity between an indi-

vidual’s current knowledge and their desired level of under-

standing. This information gap generates a state of

deprivation and creates a necessity for learning to reduce

the gap (Markey & Loewenstein, 2014). Other psychologists

such as Day (1982) suggested that the information gap needs

to be optimally sized to result in curiosity. Too large a gap

may cause anxiety, and too short a gap may not be sufﬁcient

to trigger the need for information-seeking behaviours. As we

shall demonstrate in this review, the information gap theory

repeatedly appears in studies of curiosity across disciplines.

Across a historical literature base, curiosity has mainly

been studied from the standpoint of adult humans and their

desire to ‘know’(see overview by Benedict, 2001), and

various scientiﬁc disciplines or sub-ﬁelds have focused on

categorization of curiosity rather than taking a wide view

(for example by addressing Tinbergen’s four questions;

Tinbergen, 1963). For example, psychologists and neurosci-

entists appear to have emphasized the underlying motiva-

tional processes of curiosity, whereas ethologists and

computational scientists have focused more on novelty-

seeking drivers.

Alongside the challenge of deﬁning curiosity and the lack

of a theoretical framework to study it, several methodolog-

ical posits in the ﬁeld of comparative psychology have lim-

ited the empirical investigation of curiosity in animals

(see Wynne, 2004;Shettleworth,2012). Some methodolog-

ical issues have been explored extensively in the philosoph-

ical literature [including Ockham’s Razor (see Kelly, 2004;

Shettleworth, 2010; Halina, 2015;Sober,2015;Dacey,2017;

Andrews, 2020)], with anti-anthropocentrism and Morgan’s

Canon most pertinent to our review because they have com-

monly been used as pretences to block the ascription of ‘higher’

psychological predicates such as curiosity to other animals.

In brief, anti-anthropocentrism is the principle whereby

researchers ‘reject the attribution of human traits to other animals,

usually with the implication it is done without solid justiﬁcation’

(Shettleworth, 2010, p. 477). Indeed, it is widely agreed

upon by biologists that anthropomorphizing animals does

not make for good science. However, in certain instances

it could produce an opposite problem in comparative

psychology: a ‘neganthropomorphic fallacy’(Levin, 2019).

If we deny the ascription of cognitive and psychological

predicates to systems that might genuinely possess them,

we may reinforce anthropocentrism or human exceptional-

ism. Instead, it has been argued that we should acknowledge

the phylogenetically rich interconnections that exist physio-

logically and psychologically (Levin, 2019).

The most parsimonious explanation for animal minds has

become known as ‘Morgan’s Canon’, which states that no

animal activity should be interpreted as higher psychological

processes, if it can be interpreted instead by processes which

are seen as lower in the scale of psychological evolution

(Morgan, 1903). Indeed, Morgan’s Canon is now ubiquitous

in the ﬁeld of comparative psychology, and many anecdotes

and stories are told to discourage early scholars from engag-

ing in the reasoning that leads to positing more psychological

richness in animals than is ‘actually’present. According to

Morgan’s Canon, comparative psychologists should avoid

using rich psychological terms such as ‘thinking’,‘reason-

ing’,‘evaluating’, and indeed ‘being curious’and instead

use simpler ones.

We argue that the tendency to exclude (indeed, actively

prohibit) rich psychological terms has come to impede, not

progress, the ﬁeld of comparative psychology and especially

Transdisciplinary curiosity 3

hinders comparative approaches beyond our own species.

While Morgan’sCanonhasbeeninﬂuential, a growing chorus

of science philosophers and theoreticians have called for the

jettisoning of the principle (see Sober, 2015;Andrews,2020)

because it can oversimplify our understanding of animal psy-

chology and inhibit scientiﬁc progress (Mikhalevich,

Powell & Logan, 2017; Andrews, 2020). In a similar manner

to how the study of animal emotions has recently begun to

shed light on animal psychology and deep evolutionary con-

vergences, we predict that loosening the broad prohibition

on ‘animal curiosity’will open new avenues for future

research. Indeed, in the scientiﬁc study of animal minds it is

important to have a concerted and self-conscious effort to

identify biases and prejudices that might negatively inﬂuence

experimental paradigms and scientiﬁc interpretation, but this

is a requirement of all empirical science. We thus argue that

should advance the study of animal curiosity without becom-

ing overly concerned with violating Morgan’sCanon.

III. CURIOSITY BEYOND LINGUISTIC HUMANS

In this section we discuss the different viewpoints on curiosity

where it has been studied in non-linguistic agents: animals,

human pre-verbal infants and artiﬁcial intelligences. This

body of research thus derives from interdisciplinary crossroads

between developmental psychology, neuroscience, animal

cognition and computational cognition. We will thereafter

debate in what ways we can identify similarities and discrepan-

cies across systems and propose future frameworks moving the

ﬁeld beyond disciplinary boundaries to advance our scientiﬁc

understanding of curiosity across intelligent agents.

(1) Curiosity in animals

Just like humans, animals explore their environment

(and components therein) for the sake of the activity itself

(Byrne, 2013), and as described earlier, animal studies

formed the basis for later research of curiosity in humans,

helping to shape multiple early human theories

(see Section II). It therefore seems ironic that over time,

human and animal studies of curiosity have diverged and

perhaps in doing so have moved away from meaningful

evolutionary comparisons. This is the case despite the fact

that ethologists, behavioural ecologists and zoologists have

long measured behaviours potentially interlinked with

curiosity, such as neophilia (attraction to novelty)

and exploration tendency (Glickman & Sroges, 1966;

Heinrich, 1995; Greenberg, 2003;Kendal,Coe&

Laland, 2005; Bergman & Kitchen, 2009; Mettke-

Hofmann, 2014;Forsset al., 2015;Grifﬁn, Netto &

Peneaux, 2017;Dameriuset al., 2017b).

Because researchers have been tentative in their use of the

term curiosity due to the risk of anthropomorphism (see

Section II) and the challenge empirically to assign intrinsic

motivation to animals, the operating deﬁnition is kept broad:

the motivation to seek information and learn about something unfamiliar

(Table 1). This deﬁnition derives from psychology and

Berlyne’s(

1966) work on perceptual curiosity and novel

information-seeking in rats. To underline that curious behav-

iour diverges from other activities, which directly answer bio-

logical needs, some behavioural scientists have imposed a

further criterion for an exploratory behaviour to classify as

curiosity: exploration and novelty-seeking needs to be outside

the context of general survival activities, such as feeding or searching

for protection or mates (Byrne, 2013; Kidd & Hayden, 2015).

Considering the Darwinian position that shared evolu-

tionary history incorporates the heritability of psychological

traits and the fact that behaviours interlinked with curiosity

have been demonstrated across all major animal taxa [ﬁsh

(Bisazza, Lippolis & Vallortigara, 2001;Martinset al., 2012),

amphibians (Carlson & Langkilde, 2013; Kelleher, Silla &

Byrne, 2018), reptiles (Bashaw et al., 2016; Siviter et al., 2017),

various mammals (Bergman & Kitchen, 2009;Blaser&

Heyser, 2015;Carteret al., 2018;Powellet al., 2004) and birds

Table 1. Common deﬁnitions of curiosity across scientiﬁc disciplines.

Deﬁnition of curiosity Suggested functions References Scientiﬁc discipline

A state of alert wakefulness (in the

human infant)

Deep attention paid to the

wider environment

Beiser (1984) Developmental psychology

An improvement of prediction for

getting knowledge or reduction

in uncertainty

Reducing uncertainty and

seeking information

Schmidhuber (1991b); Oudeyer &

Kaplan (2007); Friston et al.

(2017); Kidd et al.(2012)

Neuroscience

The motivation to seek

information about something

unfamiliar (beyond general

survival activities)

Motivates latent learning Berlyne (1950); Loewenstein

(1994); Byrne (2013); Damerius

et al.(

2017a); Wang & Hayden

(2021); Forss et al.(2022)

Animal cognition

The intrinsic motivation to

explore autonomously and

continuously novel space-

states/goals where the learning

progress can be maximized

over time

Motivates lifelong learning

of new information and

skills

Barto (2013); Lungarella et al.

(2003); Oudeyer & Kaplan

(2009); Santucci et al.(2020);

Schillaci et al.(

2020)

Computational modelling

4Soﬁa Forss and others

(Huber, Rechberger & Taborsky, 2001; Mettke-Hofmann

et al., 2006;O’Hara et al., 2017)], an evolutionary continuity

of curiosity becomes undeniable. Moreover, since information

processing is central to all cognitive abilities, and animal minds

represent biological information-processing devices, identifying

curiosityinanimalscanrevealinwhatwaytheyperceiveand

are motivated to process new information. There have been

three broad approaches to studying curiosity in animals which

we shall detail here: (i) exploration of a benign novel object or

novel stimulus; (ii)exploringaspeciﬁc problem/task; and (iii)

responses to abstract information.

(a)Novel object/stimulus paradigm

Because we cannot ask animals how curious they feel

(in contrast to self-reports of curiosity in humans; see Gross,

Zedelius & Schooler, 2020), behaviour is the most common

proxy for curiosity in animals. More speciﬁcally, researchers

have often experimentally induced curiosity by introducing a

novel object (or other stimulus like a scent or sound) into the

animal’s familiar environment to infer motivation to explore

and gather information about something previously

unknown (Glickman & Sroges, 1966; Bacon, 1980; Damerius

et al., 2017b; Hall et al., 2018; Forss et al., 2022; Forss &

Willems, 2022). In most cases, the novel object is inert; while

it can be manipulated and perhaps destroyed, it fundamen-

tally differs from ‘puzzles’which are task-based and can be solved

with or without retaining a food reward (see Section III.1.b).

Traditionally, the animal’s latency (i.e. duration of time)

to approach a novel object is taken as a relative measure

of their attraction towards novelty (neophilia) (Day

et al., 2003; Greenberg, 2003; Kaulfuß & Mills, 2008; Inzani,

Kelley & Boogert, 2022). It is of course possible an animal

will instead avoid the novel object, a behavioural response

referred to as neophobia (Greenberg, 1990; Fox &

Millam, 2007; Greggor et al., 2016; Forss et al., 2015; Miller

et al., 2022; Szabo & Ringler, 2023).

Critics of the novel object paradigm argue that approach-

ing a novel object does not demonstrate that an animal

obtained or learned any new knowledge (Bevins &

Besheer, 2006; Takola et al., 2021), hence does not fulﬁl some

of the deﬁnitions of curiosity imposed by some disciplinary

deﬁnitions and not seeking out novelty (i.e. retreating from

a novel object or remaining neutral towards it) cannot rule

out an animal’s potential state of curiosity even if it refrains

from physical exploration. Therefore, some scholars have

argued that for animals, especially wild ones, whose environ-

ments pose existential risks, the degree an animal can display

curiosity is interconnected with its levels of neophobia

(Greenberg, 2003; Mettke-Hofmann, 2014; Forss, Koski &

van Schaik, 2017). As such, an animal can possess both high

levels of intrinsic neophobia, serving as a protection mecha-

nism to avoid danger, and simultaneously have a high intrin-

sic motivation to explore (Reader, 2015; Moretti et al., 2015;

Forss et al., 2017). In other words, for animals, expression of

curiosity may be interlinked with their capacity to overcome

initial neophobia (Forss et al., 2017; Forss & Willems, 2022).

Consequently, we argue that risk levels in a species’

habitat (synonymous with ‘harshness’; Roth, LaDage &

Pravosudov, 2010) may have played a central role in the

evolution of curiosity (see Section III.1.d).

Beyond solely considering initial motivation to approach

and seek out novelty (neophobia/ neophilia), how long and

in what way animals explore novel objects in terms of the

diversity of manipulation actions, duration, and persistence

to explore, presents another axis of how to measure ‘observ-

able curiosity’(Pisula, 2020; Damerius et al., 2017b; Forss

et al., 2022; Birchmeier et al., 2023). As such, object manipu-

lation has the potential to deliver more descriptive data to

our understanding of their motivation to gather new infor-

mation. For example, meerkats (Suricata suricatta) explored

low-odour novel items more than novel items emitting an

odour (Birchmeier et al., 2023). This supports the idea that

the animals behaved to gain information about the novel

cue, since in situations when they cannot receive enough

information through their main sensory modality of olfac-

tion, they proceeded with physical object exploration to dis-

cover properties of the novel cue. Thus, when studying

curiosity in animals we need to consider species-speciﬁc,

evolved sensory channels through which environmental stim-

uli are processed.

A recent study, which aimed at teasing apart separate

behavioural indicators of curiosity and exploration in bottle-

nose dolphins (Tursiops truncatus) and European starlings

(Sturnus vulgaris), may offer new methodological insights to

the classic novel object paradigm (Hausberger et al., 2021).

The authors presented these species with species-speciﬁc

stimuli (for dolphins, photographs of unfamiliar dolphins or

plain water as a control; for starlings, the songs of unfamiliar

starlings and of humpback whales) and observed a period of

intense looking towards the stimuli before movement towards

them. Thus, Hausberger et al.(

2021) argued that a period

of curiosity (looking) came before exploration (moving

towards). If this is the case and can be replicated in other spe-

cies, this methodology may complement the classic novel

object test by capturing a precursive period of curiosity.

(b)Problem-solving paradigm

Despite great efforts to uncover various animal species’cog-

nitive capacities to solve problems and innovate, curiosity

has been overlooked within this body of the animal cognition

literature. From the small amount of research thus far,

orangutans (Pongo abelii and Pongo pygmaeus) that scored as

more curious (assessed through multiple novel stimulus para-

digms) and were more human oriented also used more

diverse exploratory actions and were better at solving physi-

cal cognition problems (Damerius et al., 2017a,b). This could

indicate orangutans possess a ‘curiosity drive’related to

physical cognitive challenges. Earlier work on primates

showed that rhesus macaques (Macaca mulatta) manipulate

task-based objects with no food reward, for the apparent sake

of learning how to complete the task (Davis, Settlage &

Harlow, 1950; Butler, 1954). These types of classical

Transdisciplinary curiosity 5

cognitive laboratory study do not ﬁt neatly into a novel object

nor abstract information paradigm, and so we have placed

them as a problem-solving paradigm. In studies of this kind,

animals very often invest time to ‘work’on a puzzle without

immediate reward. There are also other studies of contra

freeloading (working for food when free food is simulta-

neously available) performed with animals across laboratory,

farm, sanctuary, and zoo settings that we will not discuss

further here [for work for food despite its free availability

see Inglis, Forkman & Lazarus (1997); for work as reward

see Franks (2019)].

(c)Abstract information paradigm

Thus far, we have discussed how curiosity has been specula-

tively induced in animals by providing them with novel

objects or speciﬁc problem-solving puzzles. A limitation of

these classical paradigms is trying to demonstrate that the

demand for information scales with the amount of informa-

tion available, and it has been challenging to link knowledge

gain to a curiosity-driven search for information. In addition,

the information an animal gains from being curious may lead

to signiﬁcant future rewards, so is not truly unrewarding.

However, this problem is not unique to animals; curiosity

may potentially drive infants to engage with objects in their

environment in the absence of an immediate reward but

there is a longer-term payoff to this knowledge. As such,

exploratory actions can be argued to be evolutionary adapta-

tions for learning progress.

A recent empirical study has convincingly demonstrated

that primates will sacriﬁce a current reward to gather new

information (Wang & Hayden, 2019). This has been

achieved by giving rhesus macaques a computerized gam-

bling task and testing how much value they placed on ﬁnding

out if they had gambled correctly (Wang & Hayden, 2019).

The design was similar to humans paying for answers to quiz

questions; knowing the answers post hoc satisﬁes the subjective

feeling of human curiosity (Friston et al., 2017; Noordewier &

van Dijk, 2017). Macaques sacriﬁced a water reward to ﬁnd

out what would have happened if they gambled correctly,

even though they could not use this information to change

the outcome. Because the animals tested in this paradigm

passed the criteria of (i) sacriﬁcing reward to obtain informa-

tion, (ii) obtaining information that led to no immediate

beneﬁt, and (iii) sacriﬁcing reward proportional to the

amount of information available, Wang & Hayden (2019)

argued this was a demonstration of non-human primates

possessing human-like curiosity.

(d)Animal curiosity, ecology, and habitat

Ecology, especially in form of habitat risk, is predicted to

impact the level of curiosity both among and within species,

and so research on animal curiosity faces the challenge of

vast variation between different environments (Mettke-

Hofmann, 2014; Forss et al., 2015; Barrett, Stanton &

Benson-Amram, 2019; Inzani et al., 2022). In theory, any

animal should beneﬁt from gathering new information

regarding food sources, social partners, predators, and so

on; one would therefore expect natural selection to favour a

mechanism, like curiosity, to maximize learning from one’s

environment (Byrne, 2013; Forss et al., 2017). However, the

great risks associated with natural environments constrain

selection on curiosity, as wild animals must be vigilant for

predators and rivals and thus engaging in time-intensive

exploration is a relatively costly activity that must be

balanced with key survival behaviours like foraging (Forss

et al., 2015; van Schaik et al., 2016).

One of the most extreme environmental discrepancies

scientists have quantiﬁed is the difference between the wild

and captivity (Clark et al., 2023). Unlike wild conspeciﬁcs,

captive animals experience less-hazardous environments

and intriguingly, captivity appears to boost curiosity in mam-

mals and birds [orangutans (Forss et al., 2015; Damerius

et al., 2017b), vervet monkeys (Chlorocebus pygerythrus)

(Forss et al., 2022), spotted hyaenas (Crocuta crocuta) (Benson-

Amram, Weldele & Holekamp, 2013), rats (Rattus norvegicus),

grey short-tailed opossums (Monodelphis domestica) (Pisula

et al., 2012), meerkats (Birchmeier et al., 2023) and Gofﬁn’s

cockatoos (Cacatua gofﬁniana) (Rössler et al., 2020)]. Such a

captivity effect seems counterintuitive, given that wild ani-

mals presumably have a greater need to seek new informa-

tion than captive animals. Perhaps this phenomenon is due

to captive animals having relatively more time to spend being

curious as opposed to performing essential survival activities

(Kummer & Goodall, 1985). Connected to this is a reduction

in cognitive load and distress due to a risk-free and less

survival-focused life in captive settings. Even among free-

ranging animals that are exposed to humans, habituation

levels and repeated experience with man-made, artiﬁcial

artifacts can additionally trigger curiosity in animals (van de

Waal & Bshary, 2011; Forss et al., 2022).

From what we currently know from psychology, responses

to the stimuli eliciting human curiosity follow a U-shaped

curve, with a peak for stimuli of intermediate familiarity

(see Section III.2.a; for a review see Kidd & Hayden, 2015).

This suggests that a (human) subject possesses some back-

ground experience against which presented stimuli are eval-

uated, and accordingly triggers variations in response. If we

assume that animal curiosity follows same trajectory as

humans, animals from different habitats (such as the contrast

between wild and captive) will possess very different refer-

ence points on novelty, due to their increased experience

with human-made materials and artifacts compared to wild

individuals, and thus this experience effect may explain part

of the differences in intraspeciﬁc variation in curiosity

observed between environments. As such, cognitive ecolo-

gists uncovering how different socio-ecological conditions

inﬂuence how animals differ in their perception of their envi-

ronment can help us to understand the evolutionary path-

ways to curiosity. In the case of captive great apes, within

the same species, exposure to humans and varying degrees

of enculturation (i.e. strong human cultural inﬂuences during

development) seem to inﬂuence an individual’s levels of

6Soﬁa Forss and others

curiosity (Bering, 2004; Tomasello & Call, 2004; Russell

et al., 2011; Damerius et al., 2017b). Even though such claims

remain to be probed thoroughly, they provide an intriguing

perspective of an evolutionary continuum of curiosity and

its social and cultural inﬂuences during development.

(e)Neuroscientiﬁc research on animals

Neuroscientiﬁc work on animal curiosity (or at least linked to

curiosity, for example, exploration) has been restricted to a

handful of models [rodents, the Aplysia sea slug, and some pri-

mates (Ferdowsian, 2011; Wendler, 2014)]. However,

research addressing the neuroscience of curiosity in non-

human animals is starting to gain pace using concepts such

as uncertainty, novelty, and reward. For example, macaques

consistently prefer to receive information in advance about

an upcoming reward (a small or large amount of water), even

when prior information does not inﬂuence the size of reward

outcome (Bromberg-Martin & Hikosaka, 2009). Another

study on macaques revealed that intrinsic motivation to seek

informational reward cues derives from both uncertainty

reduction and conditioned reinforcement independently

of offered extrinsic rewards (Daddaoua, Lopes &

Gottlieb, 2016). Taken together, these results indicate that

non-human primates prefer to resolve uncertainty when they

can, and thus support the information gap theory of curiosity

(Loewenstein, 1994).

In both humans and some animal models (e.g. rodents and

non-human primates) the dopaminergic system and prefron-

tal cortex are suggestively involved in curiosity and novelty-

seeking behaviours (Bromberg-Martin & Hikosaka, 2009;

Blanchard, Hayden & Bromberg-Martin, 2015; Marvin,

Tedeschi & Shohamy, 2020) because novel stimuli can be

used to activate midbrain dopaminergic structures in both

humans and non-human animals (Horvitz, 2000; Wittmann

et al., 2007; Laurent, 2008). In gray mouse lemurs (Microcebus

murinus), the volume of the caudate nucleus is positively corre-

lated with levels of neophilia (Fritz et al., 2020; see Table 2for

glossary) and recent ﬁndings in mice show that the activity of

glutamic acid decarboxylase 2-expressing (GAD2+) inhibi-

tory neurons in the medial zona incerta (Zlm) increase signif-

icantly during explorations of novel objects (Ahmadlou

et al., 2021). Yet, it remains unclear to what extent the same

neural mechanisms are conserved more widely across phy-

logeny (Bromberg-Martin & Hikosaka, 2009; Schultz, 2016).

Whilst studies on the neuroscience of curiosity in non-

human animals have relied on the use of extrinsic rewards

(which are more easily measurable than intrinsic rewards),

some recent research has aimed to investigate curiosity with-

out extrinsic rewards. Using longitudinal functional mag-

netic resonance imaging (fMRI) and dynamic functional

connectivity analysis, Tian, Silva & Liu (2021) identiﬁed ﬁve

overlapping brain regions (the cerebellum, the hippocampus,

and cortical areas 19DI, 25, and 46D) that activated during

curiosity states without food reward in common marmosets

(Callithrix jacchus), which the authors argued was evidence

for ‘reward-free curiosity’. Further investigations of how

internal rewards shape curious behaviours in animals, as well

as examination of the brain regions related to such intrinsic

motivations, will reveal to what extent the relationship

between curiosity-related neural markers and behavioural

measures of curiosity in other species matches that of our

own (Gottlieb, Lopes & Oudeyer, 2016).

(2) Curiosity in pre-verbal infants

Children are highly curious, which enables them to learn at

an incredibly fast pace in the ﬁrst years of life (Begus &

Southgate, 2018; Kidd & Hayden, 2015). Curiosity in chil-

dren resembles one proposal by James (see Section II): an

instinctual epistemic drive to explore new objects and attend

to surprising events. The drive to approach and explore

unknown objects is a primary motor for becoming more

knowledgeable about their environment. Thus, curiosity in

children is understood as a mental state that drives

information-seeking behaviours, and which decreases in

intensity the more the sought-after information is achieved

(Berlyne, 1962,1966; Loewenstein, 1994). As children

acquire language, they can actively ask questions as a way

to explore their environment, especially when the events they

observe contradict their expectations, which requires them to

update their knowledge. While there is strong evidence for

epistemic curiosity in children (i.e. a desire for knowledge

or intellectual information; Berlyne 1960) it has been less

clear whether pre-verbal infants possess the same essential

mental capacities. Below we present recent evidence suggest-

ing there may be metacognitive processing of one’s knowl-

edge state and exploration targeted towards the unknown

during early (i.e. pre-verbal) human development.

(a)Curiosity from a developmental perspective

Exploration of the environment during early development

can maximize knowledge acquisition. Infants display a set

of innate biases and preferences for highly informative

sources of knowledge. For instance, from birth, infants prefer

to look at faces and biologically relevant stimuli as opposed to

inanimate objects (Farroni et al., 2005; Turati et al., 2002;

Valenza et al., 1996). They also display analogous preferences

in the auditory domain for informatively rich sounds, such as

infant-directed speech (Cooper & Aslin, 1990; Fernald &

Kuhl, 1987). Extant research shows that the characteristics

of infants’social environment and their previous knowledge

impact their learning and exploration of the environment.

When raised by a female primary caregiver infants develop

a visual preference for female faces by 3 months of age

(Quinn et al., 2002). When raised in an English-speaking con-

text, infants prefer English native speakers, as opposed to

non-native speakers (Kinzler, Dupoux & Spelke, 2007). Over

the course of the ﬁrst year, these preferences are tuned to

infants’environments and to the most relevant sources of

information and are geared at maximizing information gain.

It has long been shown that the pattern of exploration by

infants is not random but rather systematically enables them

Transdisciplinary curiosity 7

to learn optimally, a concept that has inspired the computa-

tional modelling of curiosity (see Section III.3). Infants prefer

to attend to the most novel objects (Fantz, 1964) or to objects

that are the least predictable or least redundant (Botvinick,

Niv & Barto, 2009; Addyman & Mareschal, 2013). They also

direct their attention towards stimuli that have optimal levels

of complexity: a phenomenon deemed ‘the Goldilocks

effect’. When pre-verbal infants (7 and 8 months old) were

presented with visual stimuli that varied parametrically in

complexity, they spent longer looking at stimuli with an inter-

mediate level of complexity compared to stimuli that were

either very simple or very complex. This pattern of visual

attention was demonstrated with different types of auditory

and visual stimuli (Kidd, Piantadosi & Aslin, 2012) and

for individual participants (Piantadosi, Jara-Ettinger &

Gibson, 2014). The ﬁndings are interpreted to mean that

infants maximize their learning by implicitly seeking

intermediate-level sources of information, which presumably

Table 2. Glossary of terms potentially interlinked with curiosity, or the ways curiosity is measured.

Term Deﬁnition References Scientiﬁc discipline

Perceptual curiosity Interest in novel, strange or ambiguous

perceptual stimuli that motivates visual

and sensory inspection

Berlyne (1954) Psychology

Epistemic curiosity The desire for knowledge or intellectual

information (applies mainly to humans)

Berlyne (1954) Psychology

Intrinsic reward Reward associated with classical task-

directed learning, food searching, and

environmental variables

Gottlieb et al.(

2013) Psychology, neuroscience

Extrinsic reward Reward associated with internal cognitive

variables such as information-seeking,

pleasure

Gottlieb et al.(

2013) Psychology, neuroscience

Novelty seeking Enhanced speciﬁc recognition, and

exploration of novel situations, stimuli,

and new experiences

Kelley et al.(

2004); Redolat

et al.(

2009); Pisula et al.

(2013)

Psychology, animal

cognition, neuroscience

Neophilia Attraction to a food item, object, or space

because it is novel. Attraction to novelty is

measured through approaches towards

that stimulus.

Greenberg (2003); Brown

& Nemes (2008); Grifﬁn

et al.(

2017)

Animal cognition

Neophobia Avoidance or aversion of novelty, measured

through restrained/inhibited behavioural

response to a certain stimulus

Reviewed by Greggor et al.

(2015) and Forss et al.

(2017)

Animal cognition

Exploration tendency/

exploration

Information acquisition/information

gathering regarding objects, space or food

items through visual investigation,

movement, or object manipulation. The

terms are vaguely deﬁned and used

broadly without standardization across

studies and species.

Massen et al.(

2013); Forss

et al.(

2017); Rojas-

Ferrer et al.(

2020); Fantz

(1964); Baer & Kidd

(2022)

Animal cognition,

developmental

psychology

Exploration Behaviours driven by an internal motivation

to ﬁnd opportunities optimal for learning

Schillaci et al.(

2020) Developmental robotics

Intrinsic motivation An adaptive function enabling organisms or

agents to learn skills and knowledge

without the requirement to impact

homeostatic needs and/or ﬁtness directly

at the time of the learning process

Baldasarre (2011) Cognitive science

Information gain Knowledge-based and competence-based

intrinsic motivation

Mirolli & Baldassarre

(2013)

Developmental robotics

Prediction gain Exploration guided towards novel or

surprising space-states as being less

explored

Bellemare et al.(

2016) Machine learning

Prediction error Difference between what is predicted and

the actual input. Measure that signals

states of surprise, novelty, information

gain, learning progress, goals, or tasks

with the optimal complexity for learning,

and prediction gain

Santucci et al.(

2020);

Schillaci et al.(

2020);

Schmidhuber (1991b);

Stahl & Feigenson,

(2015); Sutton & Barto

(2018); Valenzo et al.

(2022)

Machine learning,

computational

reinforcement learning,

developmental robotics,

cognitive sciences

8Soﬁa Forss and others

avoids wasting cognitive resources on stimuli that are too

easy or too difﬁcult to process.

Previous theoretical frameworks of curiosity have also

highlighted other mechanisms that drive early learning, such

as the feeling of uncertainty, since it signals an opportunity to

learn and the need to close a knowledge gap (Berlyne, 1962;

Loewenstein, 1994; Gottlieb et al., 2013). When they are

uncertain, children continue to explore objects and seek

information (Baer & Kidd, 2022). Pedagogical demonstra-

tions reduce uncertainty because children assume that dem-

onstrators reveal all the relevant aspects of a novel object.

After watching a teacher’s direct (i.e. to them) or indirect

(i.e. to another child) demonstration of how a novel toy

squeaks, children are less likely to explore spontaneously

other functions of that toy than after witnessing an adult’s

instrumental action on the same toy (Bonawitz et al., 2012).

Consistent with this account, children explore novel objects

more broadly when they know that their functions are only

partially, as opposed to fully, demonstrated to them by an

adult (Gweon et al., 2014). In a word-learning scenario, 3- to

8-year-old children selectively choose items that enable them

to reduce uncertainty about novel word meanings

(Zettersten & Saffran, 2021). Similarly, younger children

between 2 and 5 years of age are more likely to seek help from

others when a novel word referent is ambiguous (Hembacher,

deMayo & Frank, 2020). Globally, the more aware of their

own uncertainty children are, the more likely they are to seek

help to close the knowledge gap (Coughlin et al., 2015).

Evaluating uncertainty relies on metacognitive abilities,

through a constant assessment and awareness of what is

known and what is not known (Goupil & Proust, 2023).

Metacognitive skills, however, have long been thought to rely

on complex reasoning and the ability to report one’s own

mental states. Until recently, pre-verbal infants, and even

children, were presumed to lack metacognition, and by

extension, to lack adult forms of epistemic curiosity. In adults,

curiosity and the feeling of uncertainty are most often mea-

sured through verbal reports. However, paradigms used in

adults are inappropriate for testing children and fail to cap-

ture their metacognitive abilities. For example, when using

verbal responses children appear unaware of what they know

and what they do not know (Taylor, Esbensen &

Bennett, 1994). However, when verbal responses are

replaced with pictorial scales, preschoolers display awareness

of their uncertainty: their level of conﬁdence in their

responses reﬂects their level of accuracy (Lyons &

Ghetti, 2011). Recent studies have shown that pre-verbal

infants are also aware of gaps in their knowledge.

20-month-old toddlers can monitor their own uncertainty

and selectively request help from knowledgeable social part-

ners to avoid making errors (Goupil, Romand-Monnier &

Kouider, 2016; Bazhydai, Westermann & Parise, 2020).

12- and 18-month-old infants are also shown actively to track

the accuracy of their choices, using implicit behavioural

responses and neural measures (Goupil & Kouider, 2016).

Moreover, infants’curiosity is linked to their inner motiva-

tion to understand how the world around them works.

Recent studies in developmental psychology revealed that

infants look longer at surprising events, they systematically

explore their surroundings to learn and actively test hypoth-

eses to understand how the world around them functions

(Gopnik, Meltzoff & Kuhl, 1999; Bonawitz et al., 2012). For

instance, infants look longer at unexpected events, such as a

physical object passing through a wall. The proposed expla-

nation for this increase in attention is that surprising events

represent opportunities to learn. Decades on infant research

consistently showed increased attention following violations of

expectations in a variety of domains ranging from physical

knowledge (e.g. Baillargeon, 2008), numerical knowledge

(e.g. McCrink & Wynn, 2004;Wynn,1992), probabilistic intu-

itions (e.g. Téglas et al., 2007), logical reasoning (e.g. Cesana-

Arlotti et al., 2018) and theory of mind (e.g. Gergely et al., 1995).

Through unexpected events infants seek and attend to

sources of information to test and to reﬁne internal models

of the world that allow them to upgrade their beliefs

(Grifﬁths & Tenenbaum, 2007; Leslie, Friedman &

German, 2004). Stahl & Feigenson (2015) showed that

infants use violations of expectations of their core knowledge

as opportunities for learning. They show enhanced explora-

tion of objects that violated expectations (e.g. witnessing a

toy going through a wall, violating their expectation of object

solidity) and they engage in behaviours aimed at uncovering

an explanation for the violation (e.g. banging the toy against

a hard surface; Stahl & Feigenson, 2015). When infants are

given an explanation for the violation of expectation previ-

ously experienced (e.g. they are shown that the wall had a

hole) they no longer show increased exploratory behaviour,

suggesting that they were searching for an explanation

(Perez & Feigenson, 2022). Similarly, 6.5- to 8-month-olds

use expectation violations to revise and learn rules about

physical knowledge (Wang, Zhang & Baillargeon, 2016).

Improbable rather than impossible events, that violate non-

core knowledge, also enhance infants’exploratory behav-

iour. Using traditional violation-of-expectation methods

and crawling paradigms, it has been shown that 13-month-

olds preferentially attend to and approach sources of unex-

pected events (Sim & Xu, 2017,2019). Studies in children also

suggest that they look longer following surprising events to

understand the source of this unexpected event. For instance,

when children are presented with two confounded sources of

noise for a novel toy, their exploration is aimed at disambigu-

ating its causal mechanism (Schulz & Bonawitz, 2007).

Although on average infants look longer at violation-

of-expectation events, there is great individual variability

(Wang, Baillargeon & Brueckner, 2004;Luo&

Baillargeon, 2005). Despite this intra-individual variabil-

ity, a recent study showed that infants’attention to surpris-

ing events is quite stable within an individual: looking

behaviour at impossible events at 11 months predicts look-

ing behaviours at 17 months, but also parents’ratings of

their children’s curiosity at age 3 years (Perez &

Feigenson, 2021), even when controlling for temperamen-

tal differences, vocabulary size or attention to possible

events (Lee et al., 2023). A recent study revealed that only

Transdisciplinary curiosity 9

certain aspects of infant curiosity (i.e. broad exploration)

and of caregiver behaviour (i.e. awe-inducing activities,

such as museum visits or nature walks) were related to

early attention to surprising events (Lee et al., 2023).

Infants not only use their own surprise to learn about

unexpected events, but they are also sensitive to others’sur-

prise and use this to modify their beliefs about the world.

When 12- to 18-month-old infants witness someone else’s

surprise to improbable and probable events they show

increased looking time to the probable event (Wu &

Gweon, 2019). Similarly, children show increased explora-

tion of objects that others were surprised by during play

(Wu & Gweon, 2019).

Even before they can speak, infants learn about novel or

surprising events in their environment not only by gazing at

or manually exploring objects, but often rather than search-

ing for an explanation for themselves, they solicit one from

their caregivers. For instance, pre-verbal infants use pointing

to request information. Toddlers show superior learning of

novel labels and functions of objects they had previously

pointed to, suggesting that infant pointing is a sign of their

readiness and interest to acquire novel information

(Begus & Southgate, 2012; Lucca & Wilbourn, 2018,2019;

Lucca, 2020). When demonstrated the functions of two novel

objects, one that the infant pointed to and an ignored one,

infants selectively imitated the functions of their preferred

objects. This suggests that infants’learning is boosted when

their request is satisﬁed and that their desire to learn facili-

tates the retention of information (Begus, Gliga &

Southgate, 2014). Infants use non-verbal cues actively to

request information by selecting knowledgeable social part-

ners, as opposed to ignorant ones (Harris & Lane, 2014;

Bashydai et al., 2020; Kovacs et al., 2014). Once children

can speak, their question-asking behaviours become more

sophisticated and precise (Ronfard et al., 2018). For example,

‘why’questions are abundantly used by children to seek fur-

ther information, to understand how objects function and to

test the reliability of the information (Frazier, Gelman &

Wellman, 2009).

(b)Neuroscientiﬁc research on infants

Neuroscientists characterize curiosity as an active ‘desire’to

reduce uncertainty and thereby gather knowledge or under-

standing about novel or challenging stimuli or situations

(Gruber, Gelman & Ranganath, 2014; Silvia, 2008), thus

aligning with the information gap theory which posits that

a degree of uncertainty can intensify curiosity (Golman &

Loewenstein, 2018).

Several factors can evoke curiosity and both intrinsic and

extrinsic rewards play important roles in promoting

and maintaining high levels of curiosity in early life. Recent

studies have shown that intrinsic rewards, such as the satisfac-

tion of solving a problem or completing a task, can activate

brain regions related to high curiosity states in human infants

and children. For example, exposure to strong and repetitive

auditory stimuli in 5–7-month-old infants results in

heightened activity in the temporal lobe, which is linked to

attention, memory, and ultimately, curiosity (Emberson

et al., 2017). Furthermore, the prefrontal cortex is crucial

for attentional control, exploration, and processing of sen-

sory information in infants and this brain region has been

associated with perceptual curiosity, as it undergoes signiﬁ-

cant development during the early years of life (Gruber &

Fandakova, 2021). 8-month-old babies who were presented

with novel stimuli showed greater activity in the prefrontal

cortex, which also suggests behavioural support for this

assumption (Werchan et al., 2016). Moreover, 4-month-old

babies with strong reactions to novel and unfamiliar stimuli

showed an increase in the thickness of their cortex, speciﬁ-

cally in the region of the right ventromedial prefrontal cortex

(Schwartz et al., 2010). Another study that used functional

near-infrared spectroscopy (fNIRS) to measure brain activity

found that 3-month-old babies’bilateral prefrontal regions

show highly signiﬁcant activation towards a novel stimulus

(Nakano et al., 2009). Taken together, these results suggest

that novel and uncertain stimuli intensify curiosity states

and the speciﬁc brain region of the prefrontal cortex repre-

sents a neural marker of curiosity in human infants.

Other ﬁndings suggest the anterior cingulate cortex

(ACC), another brain region known to be involved in cogni-

tive control and decision-making, is activated during explo-

ration of novel stimuli, but not by familiarity exploration

(Bush, Luu & Posner, 2000). Also, the hippocampus, which

is involved in memory formation and retrieval, can detect

and respond to novel stimuli (Langston et al., 2010; Kidd &

Hayden, 2015). The hippocampus and the neighbouring

areas around the medial temporal lobe have been linked to

novelty preference, which in turn is believed to be crucial

for visual attention and object recognition memory necessary

for curiosity during infancy (Reynolds, 2015). Thus, these

studies imply that the ACC and hippocampus potentially

also represent neural markers of curiosity. As children grow

older, the amygdala, a region involved in processing emo-

tions and novelty detection, has been associated with curios-

ity (Jepma et al., 2012) and activity in the dopaminergic

reward system, including the ventral striatum and the pre-

frontal cortex (Kidd & Hayden, 2015; De Pisapia, Bacci &

Melcher, 2016).

(3) Computational modelling of curiosity in

artiﬁcial systems

The capability to learn new information and skills autono-

mously and continuously is strongly related to curiosity.

Therefore, the computational modelling of curiosity-

related behaviours has evoked particular interest in artiﬁcial

intelligence (Baldassarre & Mirolli, 2013), as well as allied

ﬁelds such as computational reinforcement learning

(Sutton & Barto, 2018), deep reinforcement learning

(Kulkarni et al., 2016), cognitive robotics (Schillaci,

Ciria & Lara, 2020), and developmental robotics

(Oudeyer, Kaplan & Hafner, 2007), among others. There

are multiple approaches to building artiﬁcial systems with

10 Soﬁa Forss and others

the capability autonomously to seek novel situations in the

environment and continuously to acquire new information

and skills. The computational modelling of intrinsic motiva-

tion has not only been used to permit exploration and

goal-directed learning in artiﬁcial systems (Santucci,

Baldassare & Mirolli, 2014), it has also been used for the

autonomous selection of goals or tasks (Baranes &

Oudeyer, 2013; Santucci, Baldassare & Cartoni, 2019;

Schillaci et al., 2020).

Within the literature on computational modelling of

curiosity, the mechanism behind the capability for auton-

omous open-ended learning is traditionally referred to as

intrinsic motivation. However, given that intrinsic motiva-

tion is closely related to curiosity-related behaviours, the

terms ‘intrinsic motivation’,and‘curiosity’are often used

interchangeably. Additionally, the nature of these deﬁni-

tions tends to be inﬂuenced by the discipline that a speciﬁc

study took as inspiration. This is particularly problematic

as there is no consensus on a deﬁnition in a particular

research ﬁeld and even less across research ﬁelds. Thus,

in different computational approaches, the term ‘intrinsic

motivation’has been used while disregarding what it

means in studies of psychology (for example, see

Oudeyer & Kaplan, 2009).

(a)Deﬁnitions and typologies in computational modelling of curiosity

In general, artiﬁcial systems that can autonomously explore

and behave to learn and acquire knowledge are considered

to have artiﬁcial curiosity (Schmidhuber, 1991a,b). Schillaci

et al.(

2020) proposed that intrinsic motivation exists when a

system autonomously seeks new experiences and generates

exploratory behaviours for learning by self-generating goals

depending on its current knowledge and capabilities. More

speciﬁcally related to curiosity, Oudeyer et al.(

2007), devel-

oped what they called an ‘Intelligent Adaptive Curiosity’

(IAC) system, inspired by Berlyne’s(

1960) original deﬁnitions

of intrinsic motivation in animals. The system is intelligent

because it avoids situations that are either too predictable

or too unpredictable, thus learning from situations with opti-

mal complexity (see similarities with the Goldilocks effect,

Section III.2.a). The system is adaptive because the novel sit-

uations it is attracted towards change as a function of its

learning progress. And the system has intrinsic motivation

because the learning progress, kept at its maximal level,

actively drives the system towards novel situations.

In both psychological and computational models of

intrinsic motivation, the same relevant distinction has been

made between ‘knowledge-based’and ‘competence-based’

intrinsic motivation (Mirolli & Baldassarre, 2013).

Knowledge-based intrinsic motivation is intended for accu-

mulating knowledge based on the capacity of the system to

model itself and its environment to predict the consequences

of its actions; in other words what Goupil & Proust (2023)

describe as foundations for metacognition. On the contrary,

competence-based intrinsic motivation is for acquiring skills

based on what the system can do by means of using

predictions to obtain a measure of its competence to produce

effective interactions with the environment.

Along this same line of thought, Oudeyer & Kaplan (2009)

set the ground for an operational study of intrinsic motiva-

tion by establishing a typology of different computational

approaches and distinguishing between knowledge-based

and competence-based intrinsic motivation. Knowledge-

based models of intrinsic motivation are classiﬁed into two

sub-approaches related to the way knowledge and expecta-

tions are represented: ﬁrst, information theoretic and distri-

butional models which include uncertainty motivation,

information gain motivation, distributional surprise motiva-

tion, and distributional familiarity motivation. Secondly,

predictive models include predictive novelty motivation,

intermediate level of novelty motivation, learning progress

motivation, predictive surprise motivation, and predictive

familiarity motivation. Competence-based models of intrin-

sic motivation are scarce but still some classiﬁcation has been

made to consider different approaches: maximizing incom-

petence motivation, maximizing competence progress or

ﬂow motivation, and maximizing competence.

Despite the complexity behind ﬁnding a consensus on a

deﬁnition of curiosity and intrinsic motivation in computa-

tional modelling, it is still possible to highlight an agreement.

Here we suggest that, in the computational modelling of curi-

osity, intrinsic motivation and exploration are two sides of the

same coin. Curious artiﬁcial systems need to have an internal

drive to explore and learn. This internal drive is not directly

related to extrinsic rewards; instead, it has its roots in

acquiring either knowledge or competence (Mirolli &

Baldassarre, 2013). This drive can be understood as an

attraction to novelty, and the capability to experience both

curiosity and pleasant surprise by discovering. In the litera-

ture of the computational modelling of curiosity, this internal

drive is commonly related to the terms ‘continuous learning’

or ‘lifelong learning’(Lungarella et al., 2003; Lesort

et al., 2020; De Lange et al., 2021). Lifelong learning can be

achieved as a direct consequence of being intrinsically moti-

vated to explore new space-states. Importantly, the capability

of computational modelling to explore and prefer novel situ-

ations for learning is strongly based on prediction-based

mechanisms. States of surprise elicited by novel situations

can be directly measured by means of the amount of predic-

tion error (i.e. the difference between what is predicted and

the actual input).

(b)Prediction error: the gold standard measure

Prediction error is an extremely useful measure to quantify

novelty and learning progress, the idea being that curiosity-

related behaviours can be summarized as the exploration of

what is surprising to maximize learning progress. Thus, being

able to quantify prediction error is the gold standard measure

behind computational modelling of intrinsic motivation and

exploratory-related behaviours in artiﬁcial systems. Explor-

ing novel space-states for information gain requires the

ability to measure how much the encountered information

Transdisciplinary curiosity 11

differs from what is already learned and accurately predicted.

Therefore, terms related to prediction error such as novelty,

informativeness, and surprise, are commonly used as a

measure of information gain when intrinsic motivation is

modelled, and prediction gain when exploratory behaviours

are modelled.

(c)Computational approaches to measure curiosity

Curiosity-driven learning is a promising framework for

developing autonomous lifelong learning in machines and

artiﬁcial embodied systems in developmental robotics

(Santucci et al., 2020). When it refers to (non-embodied,

non-situated) machines, the most widely used tool for

curiosity-driven learning is computational reinforcement

learning (CRL). This is likely the case because this approach

provides learning algorithms that can be applied in a wide

range of tasks and scenarios that require knowledge-based

or competence-based intrinsic motivation. When it refers to

artiﬁcial embodied systems within the developmental robot-

ics ﬁeld, there are a wide variety of tools and methods, includ-

ing neural networks, artiﬁcial evolution, and even CRL. In

developmental robotics, curiosity-driven learning has gener-

ated increasing interest due to its crucial mechanism for

ontogeny (Oudeyer et al., 2007).

Embodied and non-embodied methods share an interest

in curiosity-driven learning and the fact that prediction error

must be quantiﬁed to (i) measure novelty, (ii) determine the

optimal space-states/goals to explore, and (iii) measure

the learning progress towards acquiring new knowledge or

skill. Among the remaining salient challenges in the compu-

tational modelling of curiosity are: the autonomous selection

of actions that will lead to learning and that are useful for

solving tasks; autonomous goal selection and the related

actions for achieving it; autonomous exploration of space-

states; and the exploitation and exploration dilemma

(Santucci et al., 2020). Additionally, a well-known challenge

related to learning multiple tasks is that of ‘catastrophic for-

getting’, which means how new knowledge interferes with

what has already been learned (Parisi et al., 2018; Xiong

et al., 2021).

Finally, one of the most important remaining challenges is

the problem of relevance and decision-making between

competing internal and external drives and goals in each sit-

uation. In biological agents, innate phylogenetically acquired

knowledge, physiological needs, and emotions are crucial for

decision-making. Although there are some theoretical pro-

posals on how to model these factors in artiﬁcial systems

(Valenzo et al., 2022), it is still an open question.

(i) Non-embodied machines. For non-situated machines, the

most common computational tool is CRL, a framework that

borrows observations from behavioural psychology, particu-

larly from operant conditioning research. Thorndike (1911)

described how biological agents are affected by new stimuli

and situations associated with responses which are reinforced

by positive and negative consequences. The rewards are

external and speciﬁc to a particular environment and

context. Thus, by means of reinforcement learning, agents

learn what to do through trial and error, and how to map sit-

uations aligning with actions to maximize the reward signal

through interactions with the environment. CRL is focused

on goal-directed learning where the most rewarding actions

must be discovered in uncertain environments (Sutton &

Barto, 2018).

Traditionally, curiosity has been modelled in artiﬁcial

systems with the assumption that novelty-based exploration

is intrinsically rewarding. CRL has been widely used to build

algorithms that incorporate internal reward functions to

drive exploratory behaviours and novelty preference

(Barto, 2013). Internal rewards can be understood as a math-

ematical value associated with learning or with the acquisi-

tion of a skill (Barto, 2013). With this view, intrinsic

motivation is related to those task-independent internal

rewards, but at the same time, the artiﬁcial system needs

autonomously to seek them to be able to solve the task

(Mirolli & Baldassarre, 2013). Schmidhuber (1991a,b) was

apparently the ﬁrst to suggest that curiosity is interlinked with

a reward that is proportional to the predictability of the task

in question, and further curiosity depends on the expected

learning progress associated with a particular action. The

CRL algorithm implemented by Schmidhuber (1991b)

learned to predict the next state, given the current state,

and the execution of an action related to the task at hand.

This algorithm was designed to select those actions, the con-

sequences of which were hard to predict, but at the same time

signalled learning and intrinsic rewards for the system. The

intrinsic rewards were complemented with extrinsic rewards

related to the performance accomplished in the task. This

CRL algorithm was able to learn autonomously to improve

its own model by exploring unpredictable action states, as

well as to maximize external rewards related to the task.

CRL learning algorithms are developed with the aim to max-

imize the sum of future external and internal rewards

(Sutton & Barto, 2018).

The use of intrinsic rewards is a powerful tool to drive

learning in artiﬁcial systems, even without the use of any

external reward (Pathak et al., 2017; Burda et al., 2018). For

example, Burda et al.(

2018) designed a CRL algorithm based

solely on intrinsic rewards to compare its performance

against CRL algorithms based on external rewards. The

CRL algorithms were tested across 54 standard benchmark

environments, including the ‘Atari game suite’. Their results

show a surprisingly high performance of the CRL algorithm

on intrinsic rewards in comparison to the manually designed

extrinsic rewards of many game environments. Besides inter-

nal rewards, which are strongly related to intrinsic motiva-

tion and information gain, there are other prediction-based

mechanisms for exploration that foster learning. For exam-

ple, the frequency or count-based exploration of space-states

is used as a measure of prediction gain by exploration

(Bellemare et al., 2016). Thus, exploration can be guided

towards those novel or surprising space-states that have been

less explored. The concept of prediction gain for exploration

can be related to Sutton’s(

1991) concept of ‘exploration

12 Soﬁa Forss and others

bonus’rewards. These are rewards an artiﬁcial agent receives

for visiting states in proportion to the temporal interval since it

previously visited that state. Interestingly, Bellemare et al.

(2016) designed a novel algorithm that used both predictions

gain for exploration, and information gain for intrinsic motiva-

tion. This algorithm was applied to 2600 Atari games and

scored higher than previously reported scores.

Finally, a CRL goal-seeking system must deal with the

exploitation/exploration dilemma. This describes the choice

between exploiting what it is already learned to obtain an

expected external reward, or being intrinsically motivated

to explore and learn new information or behaviours that

could bring favourable future outcomes such as external

rewards (Sutton & Barto, 2018; for comparison with animals

in risky environments, see Section III.1.d). Competing

choices are omnipresent in the course of decision-making

and learning. Thus, a goal-seeking system needs to be

equipped with a mechanism to determine if a situation is better

suited for exploitation or exploration. It could be that intrinsi-

cally motivated behaviour serves to generate a repertoire of

skills that are useful for extrinsically motivated tasks (Barto &

Simsek, 2005). With this view, it is important to deﬁne not only

the speciﬁc environmental context but also an agent’s internal

context (Valenzo et al., 2022;Barto&Simsek,2005). This deﬁ-

nition will create the basis for making richer forms of under-

standing intrinsic rewards (Barto & Simsek, 2005).

(ii) Embodied machines. Until this point, we have focused on

the ‘cognitive’mechanisms possessed by an artiﬁcial system;

in other words, the ‘brain’it requires to be curious. How-

ever, an artiﬁcial system also needs a body to interact with

the environment and gradually acquire knowledge and

action possibilities (Pfeifer & Iida, 2004). Embodied artiﬁcial

agents need autonomously to seek and cope with novel situa-

tions to learn from these experiences and constantly increase

their behavioural complexity. Different research ﬁelds such

as psychology, neuroscience, developmental psychology,

and animal cognition (e.g. Berlyne, 1960,1966; Deci &

Ryan, 1985; Dayan & Balleine, 2002; White, 1959) have

inspired the embodiment of intrinsically motivated open-

ended learning artiﬁcial agents.

Some milestones of embodiment include the approach of

Oudeyer et al.(

2007), where an artiﬁcial agent was able to

develop, in an open-ended manner, an IAC system that used

CRL for internal rewards, as well as other computational

tools to measure prediction error and learning progress.

The IAC system maximizes learning progress by avoiding sit-

uations that are too predictable or too unpredictable. The

learning progress motivation was coupled with a

region-splitting mechanism of behaviours and sensorimotor

categories, which was used to measure prediction error by

comparing situations in a ‘regional’manner. Each region

had an associated error allowing the system to select the

region to explore which had the optimal amount of predic-

tion error. A self-organization of a developmental trajectory

to learn new possibilities of action was achieved by maximiz-

ing internal rewards when a situation previously not mas-

tered becomes mastered with the optimal time and effort.

Inspired by Forestier et al.(

2022), Schillaci et al.(2020)

presented an intrinsic motivation architecture capable of gen-

erating autonomous exploratory and curiosity-related behav-

iours in an artiﬁcial agent. Importantly, the artiﬁcial agent

was able to self-generate and select goals but was constrained

to those that generated reducible prediction error. For this

architecture, the underlying mechanism was a sensitivity to

performance and its associated artiﬁcial emotions which were

grounded on a multilevel monitoring of prediction error

dynamics. This was considered a type of self-regulating mech-

anism associated with artiﬁcial emotions. If the system detected

that prediction error was minimized when pursuing a goal, a

positive ‘emotion’was experienced, and the system was intrin-

sically motivated to continue the goal until there was no more

error to reduce. On the contrary, when prediction error was

not reduced, that is, when the system monitored that it was fail-

ingornotimprovinginpursuingagoal,anegativeemotion

was experienced. To avoid ‘frustration’, the system tried to

improve a few more cycles, but if prediction error kept not

being reduced, the system was intrinsically motivated to aban-

don the goal and to search for a new goal with lower complex-

ity. The goals were autonomously generated and selected

according to their prediction error dynamics. Selected goals

had the optimal amount of reducible prediction error. This

strategy for goal selection provides a solution for balancing

exploitation and exploration. This architecture presents a base-

line for further understanding of the relevance of monitoring

prediction error dynamics for the computational modelling of

intrinsic motivation together with artiﬁcial emotions.

Recently, there has been a convergence between the

approaches of developmental robotics and deep reinforce-

ment learning methods for tracking the problem of intrinsic

motivation and lifelong learning. This new domain of study

is called developmental machine learning. Colas et al.(

2020)

proposed a typology of the methods to train artiﬁcial agents

to generate and pursue their own goals based on developmen-

tal machine learning. This typology considers Intrinsically

Motivated Goal Exploration Processes (IMGEPs), with

Goal-Conditioned (GC) Reinforcement Learning algorithms.

Under this view, artiﬁcial agents equipped with a

GC-IMGEPs system will autonomously represent their goals,

generate them, and will be intrinsically motivated to learn to

achieve the goal while measuring their own progress. In the

proposed typology, different types of goal representation asso-

ciated with a speciﬁc goal-conditioned reward function were

identiﬁed. Additionally, in typology the reward functions

(i.e. giving a positive reward) are explained considering which

type of goal representation is selected for measuring progress.

IV. DISCUSSION

(1) Disciplinary overlaps and their implication for

current views on curiosity

In this review, we sought to shine a spotlight on curiosity research

on non-linguistic systems: non-human animals, pre-verbal

Transdisciplinary curiosity 13

infants, and artiﬁcial agents. We were interested to reveal to

what extent these systems differ oroverlapintheirassessments

of curiosity, and how they may move the study of curiosity

away from the more dominant paradigms in human psy-

chology. This review has revealed three main ﬁndings. (i)

Across all ‘intelligent systems’curiosity can be sparked by

surprise and uncertainty and plays a central role in optimiz-

ing learning processes in the presence of novel stimuli and

situations. (ii) Across systems, the level of uncertainty needs

to be optimal to result in curiosity: too large a gap may

cause anxiety/retreat (expressed through neophobia in ani-

mals), and too short a gap may not be sufﬁcient to trigger

the need for information-seeking behaviours (Day, 1982).

(iii) The information gain itself (resulting from exploration)

is considered rewarding for the agent.

In Table 1, we summarize how different scientiﬁc disci-

plines have deﬁned curiosity. As expected, in systems where

one cannot use self-assessments nor infer a state of the mind,

deﬁnitions are broad (arguably ambiguous) but with the com-

monality of describing a motivation to aid learning processes.

Interestingly, within animal cognition, the criterion for curi-

osity is somewhat stricter (information-seeking or exploratory

behaviours must fall outside general survival activities

and/or cannot have any immediate reward to the exploring

individual) (Byrne, 2013; Wang & Hayden, 2019), whilst

for other disciplines the actual explorative action is driven

by information being the reward (Loewenstein, 1994;

Bromberg-Martin & Hikosaka, 2009; Zhai et al., 2019).

In computational cognition, the inspiration for prediction

error calculations derives from how human infants attend to

stimuli when exploring their environments. For example,

infants prefer to attend to those stimuli that are neither too

complex nor too simple, maintaining an intermediate rate

of information complexity to be able to learn incrementally

(Kidd et al., 2012). In line with the theory behind the

U-shaped curve optimizing learning progress (Kidd &

Hayden, 2015), the concept of optimal complexity in compu-

tational cognition provides a logical solution to the problem

of what is relevant to learn and what must be ignored because

it is already learned (Stahl & Feigenson, 2015). Thus, in the

same way that computational models use optimal levels of

complexity to facilitate the balance between exploration

and exploitation, such models also offer a possibility to eval-

uate whether a similar learning-optimization strategy is

applicable for both neural networks and artiﬁcial ones. To

what extent other non-human species adjust novelty interest

and learning progress according to similar reference points is

yet for animal cognition scientists to uncover.

Researchers in the ﬁeld of animal cognition have been

cautious when classifying a behaviour as curiosity because it

needs to be sufﬁciently differentiated from other concepts like

neophilia/neophobia, and there is also an ingrained concern

about anthropomorphism and violating Morgan’s Canon.

By contrast, child psychologists seem more freely able to infer

that curiosity is the driver behind exploration and learning in

babies, even though many of the behavioural indicators used

to do so are the same for animals (Section II).

In Table 2we list different terminology used across disci-

plines that interlink with curiosity. Across disciplines, multi-

ple terms describe the initial step of recognizing/seeking

new information, which is necessary to guide an organism

(or agent) towards curiosity-driven learning: ‘perceptual

curiosity’(psychology), ‘novelty-seeking’(psychology, animal

cognition, neuroscience), ‘neophilia’(animal cognition),

‘exploration’(animal cognition, developmental psychology,

developmental robotics) (Table 2).

According to the literature, the distinction between what is

information and what is knowledge colours how curiosity

is viewed. As we have seen, both biological and artiﬁcial

agents require an internal mechanism for information-

seeking in order to adapt to their environments. Such inter-

nal motivation is the basic biological mechanism underlying

exploratory behaviours, combining internal states (such as

an organism’s metabolism, hunger state or level of uncer-

tainty) with encounters with external stimuli (such as food

presence or novel stimulus) (Pisula, Turlejski &

Charles, 2013). However, true curiosity-based exploration

has been argued to require an additional element; the assess-

ment of newness followed by a pursued motivation to gain

more knowledge about the discovered available new infor-

mation (Pisula, 2020).

Beyond humans (who can self-report), curiosity is a con-

tentious phenomenon to quantify because it is difﬁcult to

measure meta-cognition in non-linguistic beings. Some cat-

egorize curiosity with affective states or epistemic emotions

as strictly ‘ﬁrst order’, just like feelings of fear, and thus do

not incorporate a monitoring function. In this view,

non-human animals and infants engage in explorative

behaviours without using or having the concept of knowl-

edge (Carruthers, 2018). However, recently an intriguing

contra-proposal was made by Goupil & Proust (2023),

who argue that curiosity is a special type of ‘metacognitive

feeling’and represents an affective state resulting from an

agent monitoring the success or failure of its own cognitive

actions,whichinthecaseofcuriosity would be identifying

and recognizing potential new information and acting

thereupon. In fact, Goupil & Proust (2023) suggest that

developing babies and animals engaging in curious

information-seeking represents rudimentary evidence for

metacognitive competence. Thus, seeking and/or recogniz-

ing new information may be a pre-condition for true knowl-

edge gain of a certain stimulus/situation.Aswehaveseen,

human infants show the same pre-condition of curiosity

from their ﬁrst months of life. Their behavioural explora-

tions, requests for help and evaluations of their own and

others’knowledge become more sophisticated across the

course of their development. By documenting cases across

non-linguistic systems showing increased interest in novelty,

exploration of new stimuli and the balance of such explor-

ative acts depending on external factors inﬂuencing assess-

ment, our review supports the provocative suggestion

made by Goupil & Proust (2023) of curiosity representing

a fundamental meta-cognitive trait, which likely stretches

(at least in some forms) beyond human adults.

14 Soﬁa Forss and others

(2) Do we need a dichotomic view of curiosity?

An early (19th century) division of curiosity into two distinct

types (biological and uniquely human/requiring language;

James, 1890) left a long-lasting impression on its scientiﬁc

study. Whilst many scholars have acknowledged the

existence of curiosity in animals and its biological function

as a mechanism necessary for learning, others have instead

argued that such explorative behaviour represents a clear-

cut distinct phenomenon to what James referred to as ‘scien-

tiﬁc curiosity’, which he described as the human-like desire to

ﬁll a knowledge gap (James, 1890). In the current literature

there are two other major divisions of curiosity which have

prevented straightforward deﬁnition: (i) a drive to seek out

information, and (ii) an underlying internal motivation to

learn something new (Marvin et al., 2020). By reviewing the

curiosity literature extending beyond adult humans, we have

highlighted that all intelligent systems, whether human

babies (unable to assess their own knowledge status), non-

human animals or artiﬁcial agents, possess the ability to seek

and recognize new information, and that this ability is facili-

tated by intrinsic motivation. Thus, one can start to question

to what extent the traditional dichotomic view is a reﬂection

of the available (or in many cases non-available) tools to mea-

sure internal motivations in non-linguistic beings, rather than

clear evidence for human uniqueness.

One way to bridge the gap between these dichotomous

divisions would be to address both views within the same

model system. While somewhat time consuming, yet highly

interesting, one could apply a longitudinal approach in

humans to investigate the relationship between behavioural

measures and later-life self-assessments within individuals

over time. Behavioural measures from animal cognition

and developmental psychology could evaluate early curious

tendencies in babies, which can be complemented with neu-

roscientiﬁc approaches to detect signals of neurological

markers of curiosity at different developmental stages. The

same individual’s behaviour and self-evaluation can be fol-

lowed up later in life to evaluate the inter-individual agree-

ment across methods and life stages.

(3) Transdisciplinary future approaches

Whilst behavioural measures of exploration represent what is

‘observable’about curiosity across animals, infants, and arti-

ﬁcial agents, capturing the essence of the trait in comparative

work remains a challenge and may require a combination of

paradigms. One obvious overlap between developmental

psychology and animal cognition is the observation of explor-

ative behaviour. Quantifying interest towards various objects

and problem-solving can be achieved using multiple beha-

vioural indicators, such as spatially approaching/retreating,

gaze direction and ﬁelds of attention, attention spans, types

of object manipulation, gesturing, and persistence. Such

multi-component measures of exploration are already estab-

lished in great ape studies and the umbrella term ‘curiosity’

has been used to combine underlying correlated exploratory

measures (Damerius et al., 2017b; Forss & Willems, 2022).

We suggest that a model of curiosity should include multiple

components, like exploration, knowledge awareness, and vio-

lation of expectation that can be comparatively studied in

non-linguistic beings. We currently know very little about

how such traits may be intercorrelated in animals and human

infants, but this knowledge is crucial to understanding the

building blocks of curiosity and its evolutionary history.

Despite their intimate relationship in understanding the part

curiosity plays in the process of learning, and the fact that

these two scientiﬁcﬁelds face similar restrictions on the use

of language-based methods, surprisingly few direct empirical

comparative studies exist between human toddlers and our

closest living relatives (but see Herrmann et al., 2011). The

comparative interest between these disciplines has been

larger for the related concept of play behaviour

(Pellegrini & Smith, 2005).Herewesuggestthatsuitable

experimental designs that examine curiosity using beha-

vioural measures open the intriguing possibility to quantify

aspects of curiosity not only in non-human animals and pre-

verbal infants, who are too young for self-reported measures

on internal states, but also to some extent in artiﬁcial intel-

ligences, whose progress also depends on curiosity-driven

learning (Baldassarre & Mirolli, 2013;Colaset al., 2020;

Oudeyer & Kaplan, 2009; Schmidhuber, 1991a,b;

Sutton & Barto, 2018).

To this end, new methods such as measuring theta brain

wave oscillations [as an indicator of active learning see

Begus & Southgate (2018) and Begus & Bonawiz (2020)]

could help to identify neural correlates of curiosity across a

phylogeny (e.g. in human infants and great apes). A more

practical method that has already been validated for

non-human use is non-invasive eye-tracking (Kano &

Tamonaga, 2009;Karlet al., 2020;Lewis&

Krupenye, 2022). The only shortcoming of such comparative

paradigms is that they require extensive training in non-

human animals, and as such can only be applied in highly con-

trolled laboratory or research centre conditions; this requires

extensive exposure to humans and thus animals may become

partly enculturated through their experiences. Given the

reported differences in curiosity between wild and captive set-

tings (Section III.1.d), such methods will perhaps have limited

ecological/biological validity.

Where technological tools may fall short in their transla-

tion to non-human animals, we still argue strongly for inten-

siﬁed dialogue between developmental psychologists and

animal behaviour researchers. As discussed herein, human

infants use pointing gestures as a request for knowledge

(Kita, 2003; Tommasello, Carpenter & Liszkowski, 2007),

and so the desire to ‘know more’is present early in human

development. When learning their diet, great ape infants

watch food items that are rare in their mother’s diet for lon-

ger (Jaeggi et al., 2010; Schuppli et al., 2016), thus indicating

an ability to recognize and seize a novel learning opportu-

nity. Moreover, ape infants also beg more frequently for food

items from their mothers that are harder to process, which

has been suggested as a request for information on how to

Transdisciplinary curiosity 15

process such items rather than a sole need for nutrition

(Jaeggi, Van Noordwijk & Van Schaik, 2008). These isolated

examples point towards phylogenetic commonalities worthy

of further investigation.

If curiosity and its functions are approached in similar

ways to the scientiﬁc study of intelligence, we may consider

taking a ‘modular’or ‘domain’approach where we measure

multiple interconnected underlying behaviours (surprise

and/or uncertainty-evoked reactions, novelty exploration,

violation of expectations, etc.) that combine to produce an

organism’s curiosity state. And in the continued spirit of tak-

ing inspiration from biological systems, we recommend that

future research on the functions of curiosity is framed around

Tinbergen’s four questions (survival value, ontogeny, evolu-

tion, and causation; Tinbergen, 1963). This will ensure that

curiosity research remains open and avoids compartmentali-

zation. In particular, we encourage more neuroscientiﬁc

research, which can make signiﬁcant contributions to our

understanding of underlying causative mechanisms like

needs, and links to adaptive functions such as enhanced

memory capacity. Finally, since curiosity is a function of lived

experience, we urge more research examining how curiosity

changes across an agent’s lifetime, from birth to death.

V. CONCLUSIONS

(1) Research approaches on non-linguistic agents (where one

cannot infer a state of the mind or use self-assessments) collec-

tively deﬁne curiosity broadly as ‘a motivation to aid learning

processes’.

(2) Throughout this review, it is clear that reducing uncer-

tainty is a common prediction of when and how curiosity is

expressed across systems.

(3) In all disciplines (developmental psychology, animal cog-

nition, neuroscience, and computational cognition) we found

approaches that are united in the view and methodology that

curiosity can be sparked by surprise and uncertainty.

(4) Our review supports the provocative suggestion made by

Goupil & Proust (2023) that curiosity represents a fundamen-

tal meta-cognitive trait that potentially stretches (at least in

some forms) beyond human adults.

(5) We welcome and outline intradisciplinary exchange not

only on theoretical levels but also regarding methodologies.

(6) We suggest that a model of curiosity should include mul-

tiple components, like exploration, knowledge awareness,

and violation of expectation that can be studied compara-

tively also in non-linguistic beings.

VI. ACKNOWLEDGEMENTS

We would like to thank Erica Cartmill and Jacob Foster from

the Diverse Intelligence Summer Institute and all our cohorts

in the edition of Summer 2021, when our intradisciplinary

work sparked. Further, we are grateful to the Collegium

Helveticum, the joint Institute for Advanced Studies of

ETH, UZH, and ZHdK in Zurich, Switzerland for ﬁnancial

contribution to our cross-disciplinary workshop which facili-

tated the work on this review. The authors declare no conﬂict

of interest.

VII. AUTHOR CONTRIBUTIONS

This review is a product of a Diverse Intelligences Summer

Institute run by UCLA followed by a transdisciplinary work-

shop, conceptualized, and organized by S. F. together with

the Collegium Helveticum, ETH Zurich. The introduction

and discussion were written by S. F. and F. C. Disciplinary

sections and tables of deﬁnitions and terminology were led

by the following authors: historical overview –D. H.; devel-

opmental psychology –C.-I. G.; animal cognition –S. F.

and F. C.; neuroscience –S. L.; computational model-

ling –A. C.

VIII. REFERENCES

Addyman, C. &Mareschal, D. (2013). Local redundancy governs infants’

spontaneous orienting to visual-temporal sequences. Child Development 84(4),

1137–1144.

Ahmadlou, M.,Houba, J. H.,van Vierbergen, J. F.,Giannouli, M.,

Gimenez, G. A.,van Weeghel, C.,Darbbanfouladi, M.,Yasamin

Shirazi, M.,Dziubek, J.,Kacem, M. d.,Winter, F. &Heimel, J. A. (2021). A

cell type–speciﬁc cortico-subcortical brain circuit for investigatory and novelty-

seeking behavior. Science 372(6543), eabe9681.

Andrews, K. (2020). The Animal Mind: An Introduction to the Philosophy of Animal Cognition.

Routledge, Abingdon.

Bacon, E. S. (1980). Curiosity in the American black bear. In Bear Biology Association

Conference Series (Volume 4), pp. 153–157. Jstore, Montana.

Baer, C. &Kidd, C. (2022). Learning with certainty in childhood. Trends in Cognitive

Sciences 26(10), 887–896.

Baillargeon, R. (2008). Innate ideas revisited: for a principle of persistence in infants’

physical reasoning. Perspectives on Psychological Science 3(1), 2–13.

Baldassarre, G. (2011). What are intrinsic motivations? A biological perspective. In

2011 IEEE International Conference on Development and Learning (ICDL), pp. 1–8. Frankf urt

am Main, Germany.

BALDASSARRE,G.&MIROLLI, M. (eds) (2013). Intrinsically Motivated Learning in Natural and

Artiﬁcial Systems, pp. 1–14. Springer, Berlin Heidelberg.

Baranes, A. &Oudeyer, P. Y. (2013). Active learning of inverse models with

intrinsically motivated goal exploration in robots. Robotics and Autonomous Systems 61,

49–73.

Barrett, L. P.,Stanton, L. A. &Benson-Amram, S. (2019). The cognition of

‘nuisance’species. Animal Behaviour 147, 167–177.

Barto, A. G. (2013). Intrinsic motivation and reinforcement learning. In Intrinsically

Motivated Learning in Natural and Artiﬁcial Systems, pp. 17–47. Springer Berlin,

Heidelberg.

Barto, A. G. &Simsek, O. (2005). Intrinsic motivation for reinforcement learning

systems. In Proceedings of the Thirteenth Yale Workshop on Adaptive and Learning Systems,

pp. 113–118. Yale University Press, New Haven, CO.

Bashaw, M. J.,Gibson, M. D.,Schowe, D. M. &Kucher, A. S. (2016). Does

enrichment improve reptile welfare? Leopard geckos (Eublepharis macularius)

respond to ﬁve types of environmental enrichment. Applied Animal Behaviour Science

184, 150–160.

Bazhydai, M.,Westermann, G. &Parise, E. (2020). ‘I don’t know but I know who

to ask’: 12-month-olds actively seek information from knowledgeable adults.

Developmental Science 23(5), e12938.

Begus, K. &Bonawitz, E. (2020). The rhythm of learning: theta oscillations as an

index of active learning in infancy. Developmental Cognitive Neuroscience 45, 100810.

Begus, K.,Gliga, T. &Southgate, V. (2014). Infants learn what they want to learn:

responding to infant pointing leads to superior learning. PLoS One 9(10), e108817.

16 Soﬁa Forss and others

Begus, K. &Southgate, V. (2012). Infant pointing serves an interrogative function.

Developmental Science 15(5), 611–617.

Begus, K. &Southgate, V. (2018). Curious learners: how infants’motivation to

learn shapes and is shaped by infants’interactions with the social world. In Active

Learning from Infancy to Childhood: Social Motivation, Cognition, and Linguistic Mechanisms,

pp. 13–37. Springer International Publishing AG, Cham.

Beiser, H. R. (1984). On curiosity: a developmental approach. Journal of the American

Academy of Child Psychiatry 23(5), 517–526.

Bellemare, M.,Srinivasan, S.,Ostrovski, G.,Schaul, T.,Saxton, D. &

Munos, R. (2016). Unifying count-based exploration and intrinsic motivation.

Advances in Neural Information Processing Systems 29, 1471–1479.

Benedict, B. M. (2001). Curiosity: A Cultural History of Early Modern Inquiry. University of

Chicago Press, Chicago.

Benson-Amaram, S.,Weldele, M. L. &Holekamp, K. E. (2013). A comparison of

innovative problem-solving abilities between wild and captive spotted hyaenas,

Crocuta crocuta.Animal Behaviour 85(2), 349–356.

Bergman, T. J. &Kitchen, D. M. (2009). Comparing responses to novel objects in

wild baboons (Papio ursinus) and geladas (Theropithecus gelada). Animal Cognition 12,

63–73.

Bering, J. M. (2004). A critical review of the ‘enculturation hypothesis’: the effects of

human rearing on great ape social cognition. Animal Cognition 7, 201–212.

Berlyne, D. E. (1950). Novelty and curiosity as determinants of exploratory

behaviour. British Journal of Psychology 41(1), 68.

Berlyne, D. E. (1954). An experimental study of human curiosity. British Journal of

Psychology 45(4), 256.

Berlyne, D. E. (1955). The arousal and satiation of perceptual curiosity in the rat.

Journal of Comparative and Physiological Psychology 48(4), 238–246.

Berlyne, D. E. (1960). Conﬂict, Arousal, and Curiosity. McGraw-Hill Book Company,

New York.

Berlyne, D. E. (1962). Uncertainty and epistemic curiosity. British Journal of Psychology

53(1), 27–34.

Berlyne, D. E. (1966). Curiosity and exploration: animals spend much of their time

seeking stimuli whose signiﬁcance raises problems for psychology. Science

153(3731), 25–33.

Bevins, R. A. &Besheer, J. (2006). Object recognition in rats and mice: a one-trial

non-matching-to-sample learning task to study recognition memory. Nature Protocols

1(3), 1306–1311.

Birchmeier, K.,Johnson-Ulrich, L.,Stein, J. &Forss, S. (2023). The role of

umwelt in animal curiosity: a within and between species comparison of novelty

exploration in mongooses. Animal Behaviour and Cognition 10(4), 329–354.

Bisazza, A.,Lippolis, G. &Vallortigara, G. (2001). Lateralization of ventral ﬁns

use during object exploration in the blue gourami (Trichogaster trichopterus).

Physiology & Behavior 72(4), 575–578.

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.

Blaser, R. &Heyser, C. (2015). Spontaneous object recognition: a promising

approach to the comparative study of memory. Frontiers in Behavioral Neuroscience

9, 183.

Bonawitz, E. B.,van Schijndel, T. J.,Friel, D. &Schulz, L. (2012). Children

balance theories and evidence in exploration, explanation, and learning. Cognitive

Psychology 64(4), 215–234.

Botvinick, M. M.,Niv, Y. &Barto, A. G. (2009). Hierarchically organized behavior

and its neural foundations: a reinforcement learning perspective. Cognition 113(3),

262–280.

Bromberg-Martin, E. S. &Hikosaka, O. (2009). Midbraindopamine neurons signal

preference for advance information about upcoming rewards. Neuron 63(1), 119–126.

Brown, G. R. &Nemes, C. (2008). The exploratory behaviour of rats in the hole-

board apparatus: is head-dipping a valid measure of neophilia? Behavioural Processes

78(3), 442–448.

Burda, Y.,Edwards, H.,Pathak, D.,Storkey, A.,Darrell, T. &Efros, A. A.

(2018). Large-scale study of curiosity-driven learning. arXiv preprint arXiv:

1808.04355.

Bush, G.,Luu, P. &Posner, M. I. (2000). Cognitive and emotional inﬂuences in

anterior cingulate cortex. Trends in Cognitive Sciences 4(6), 215–222.

Butler, R. A. (1954). Curiosity in monkeys. Scientiﬁc American 190(2), 70–75.

Byrne, R. W. (2013). Animal curiosity. Current Biology 23(11), 469–470.

Carlson, B. E. &Langkilde, T. (2013). Personality traits are expressed in bullfro g

tadpoles during open-ﬁeld trials. Journal of Herpetology 47(2), 378–383.

Carruthers, P. (2018). Basic questions. Mind & Language 33, 130–147.

Carter, A. J.,Forss, S.,Page, R. A. &Ratcliffe, J. M. (2018). Younger vampire

bats (Desmodus rotundus) are more likely than adults to explore novel objects. PLoS

One 13(5), e0196889.

Cesana-Arlotti, N.,Martin, A.,Teglas, E.,Vorobyova, L.,Cetnarski, R. &

Bonatti, L. L. (2018). Precursors of logical reasoning in preverbal human infants.

Science 359(6381), 1263–1266.

Clark, F. E.,Greggor, A. L.,Montgomery, S. H. &Plotnik, J. M. (2023). The

endangered brain: actively preserving ex-situ animal behaviour and cognition will

beneﬁt in-situ conservation. Royal Society Open Science 10(8), 230707.

Colas, C.,Karch, T.,Sigaud, O. &Oudeyer, P. Y. (2020). Intrinsically motivated

goal-conditioned reinforcement learning: a short survey. arXiv preprint arXiv:

2012.09830.

Cooper, R. P. &Aslin, R. N. (1990). Preference for infant-directed speech in the ﬁrst

month after birth. Child Development 61(5), 1584–1595.

Coughlin, C.,Hembacher, E.,Lyons, K. E. &Ghetti, S. (2015). Introspection on

uncertainty and judicious help-seeking during the preschool years. Developmental

Science 18(6), 957–971.

Dacey, M. (2017). Anthropomorphism as cognitive bias. Philosophy of Science 84(5),

1152–1164.

Daddaoua, N.,Lopes, M. &Gottlieb, J. (2016). Intrinsically motivated

oculomotor exploration guided by uncertainty reduction and conditioned

reinforcement in non-human primates. Scientiﬁc Reports 6(1), 1–15.

Damerius, L. A.,Forss, S. I. F.,Kosonen, Z. K.,Willems, E. P.,Burkart, J. M.,

Call, J.,Galdikas, B. M. F.,Liebal, K.,Haun, D. B. M. &van Schaik, C. P.

(2017a). Orientation toward humans predicts cognitive performan ce in orang-utans.

Scientiﬁc Reports 7(1), 1–12.

Damerius, L. A.,Graber, S. M.,Willems, E. P. &van Schaik, C. P. (2017b).

Curiosity boosts orang-utan problem-solving ability. Animal Behaviour 134,57–70.

Davis, R. T.,Settlage, P. H. &Harlow, H. F. (1950). Performance of normal and

brain-operated monkeys on mechanical puzzles with and without food incentive. The

Pedagogical Seminary and Journal of Genetic Psychology 77(2), 305–311.

Day, H. I. (1982). Curiosity and the interested explorer. Performance & Instruction. 21,

19–22.

Day, R. L.,Coe, R. L.,Kendal, J. R. &Laland, K. N. (2003). Neophilia, innovation

and social learning: a study of intergeneric differences in callitrichid monkeys. Animal

Behaviour 65(3), 559–571.

Dayan, P. &Balleine, B. W. (2002). Reward, motivation, and reinforcement

learning. Neuron 36, 285–298.

De Lange, M.,Aljundi, R.,Masana, M.,Parisot, S.,Jia, X.,Leonardis, A.,

Slabaugh, G. &Tuytelaars, T. (2021). A continual learning survey: defying

forgetting in classiﬁcation tasks. IEEE Transactions on Pattern Analysis and Machine

Intelligence 44(7), 3366–3385.

De Pisapia, N.,Bacci, F. &Melcher, D. (2016).Brain networks for visualcreativity: a

functional connectivity study of planning a visual artwork. ScientiﬁcReports6(1), 39185.

Dean, V.,Tulsiani, S. &Gupta, A. (2020). See, hear, explore: curiosity via audio-

visual association. Advances in Neural Information Processing Systems 33, 14961–14972.

Deci, E. L. &Ryan, R. M. (1985). Intrinsic Motivation and Self-Determination in Human

Behavior. Springer Science & Business Media, New York.

Emberson, L. L.,Cannon, G.,Palmeri, H.,Richards, J. E. &Aslin, R. N.

(2017). Using fNIRS to examine occipital and temporal responses to stimulus

repetition in young infants: evidence of selective frontal cortex involvement.

Developmental Cognitive Neuroscience 23,26–38.

Fantz, R. L. (1964). Visual experience in infants: decreased attention to familiar

patterns relative to novel ones. Science 146(3644), 668–670.

Farroni, T.,Johnson, M. H.,Menon, E.,Zulian, L.,Faraguna, D. &

Csibra, G. (2005). Newborns’preference for face-relevant stimuli: effects of

contrast polarity. Proceedings of the National Academy of Sciences 102(47), 17245–17250.

Ferdowsian, H. (2011). Human and animal research guidelines: alignin g ethical

constructs with new scientiﬁc developments. Bioethics 25(8), 472–478.

Fernald, A. &Kuhl, P. (1987). Acoustic determinants of infant preference for

motherese speech. Infant Behavior and Development 10(3), 279–293.

Forestier, S.,Portelas, R.,Mollard, Y. &Oudeyer, P. Y. (2022). Intrinsically

motivated goal exploration processes with automatic curriculum learning. The

Journal of Machine Learning Research 23(1), 6818–6858.

Forss, S. &Willems, E. (2022). The curious case of great ape curiosity and how it is

shaped by sociality. Ethology 128(8), 552–563.

Forss, S. I. F.,Koski, S. E. &van Schaik, C. P. (2017). Explaining the paradox of

neophobic explorers: the social information hypothesis. International Journal

of Primatology 38, 799–822.

Forss, S. I. F.,Motes-Rodrigo, A.,Dongre, P.,Mohr, T. &van de Waal, E.

(2022). Captivity and habituation to humans raise curiosity in vervet monkeys.

Animal Cognition 25,1–12.

Forss, S. I. F.,Schuppli, C.,Haiden, D.,Zweifel, N. &van Schaik, C. P. (2015).

Contrasting responses to novelty by wild and captive orangutans. American Journal of

Primatology 77(10), 1109–1121.

Fox, R. A. &Millam, J. R. (2007). Novelty and individual differences inﬂuence

neophobia in orange-winged Amazon parrots (Amazona amazonica). Applied Animal

Behaviour Science 104(1–2), 107–115.

Franks, B. (2019). What do animals want? Animal Welfare 28(1), 1–10.

Frazier, B. N.,Gelman, S. A. &Wellman, H. M. (2009). Preschoolers’search for

explanatory information within adult–child conversation. Child Development 80(5),

1592–1611.

Transdisciplinary curiosity 17

Friston, K. J.,Lin, M.,Frith, C. D.,Pezzulo, G.,Hobson, J. A. &Ondobaka, S.

(2017). Active inference, curiosity and insight. Neural Computation 29(10), 2633–2683.

Fritz, R. G.,Zimmermann, E.,Meier, M.,Mestre-Frances, N.,

Radespiel, U. &Schmidtke, D. (2020). Neurobiological substrates of animal

personality and cognition in a nonhuman primate (Microcebus murinus). Brain and

Behavior 10(9), e01752.

Gallagher, M. W. &Lopez, S. J. (2007). Curiosity and well-being. The Journal of

Positive Psychology 2(4), 236–248.

Gergely, G.,Nadasdy, Z.,Csibra, G. &Biro, S. (1995). Taking the intentional

stance at 12 months of age. Cognition 56(2), 165–193.

Glickman, S. E. &Hartz, K. E. (1964). Exploratory behavior in several species of

rodents. Journal of Comparative and Physiological Psychology 58(1), 101–104.

Glickman, S. E. &Sroges, R. W. (1966). Curiosity in zoo animals. Behaviour 26(1–2),

151–187.

Golman, R. &Loewenstein, G. (2018). Information gaps: a theory of preferences

regarding the presence and absence of information. Decision 5(3), 143–164.

Gopnik, A.,Meltzoff, A. N. &Kuhl, P. K. (1999). The Scientist in the Crib: Minds,

Brains, and how Children Learn. William Morrow & Co, New York.

Gottfredson, L. S. (1998). The general intelligence factor. Scientiﬁc American 9,

24–30.

Gottlieb, J.,Lopes, M. &Oudeyer, P. Y. (2016). Motivated cognition: neural and

computational mechanisms of curiosity, attention, and intrinsic motivation. In Recent

Developments in Neuroscience Research on Human Motivation. Emerald Group Publishing

Limited, Leeds.

Gottlieb, J.,Oudeyer, P. Y.,Lopes, M. &Baranes, A. (2013). Information-

seeking, curiosity, and attention: computational and neural mechanisms. Trends in

Cognitive Sciences 17(11), 585–593.

Goupil, L. &Kouider, S. (2016). Behavioral and neural indices of metacognitive

sensitivity in preverbal infants. Current Biology 26, 3038–3045.

Goupil, L. &Proust, J. (2023). Curiosity as a metacognitive feeling. Cognition 231,

105325.

Goupil, L.,Romand-Monnier, M. &Kouider, S. (2016). Infants ask for help

when they know they don’t know. Proceedings of the National Academy of Sciences

113(13), 3492–3496.

Greenberg, R. S. (1990). Ecological plasticity, neophobia, and resource use in birds.

Studies in Avian Biology 13, 431–437.

Greenberg, R. S. (2003). The role of neophobia and neophilia in the development of

innovative behaviour of birds. In Animal innovation (eds S. READER and K. LALAND).

Oxford University Press, Oxford.

Greggor, A. L.,Jolles, J. W.,Thornton, A. &Clayton, N. S. (2016). Seasonal

changes in neophobia and its consistency in rooks: the effect of novelty type and

dominance position. Animal Behaviour 121,11–20.

Greggor, A. L.,Thornton, A. &Clayton, N. S. (2015). Neophobia is not only

avoidance: improving neophobia tests by combining cognition and ecology. Current

Opinion in Behavioral Sciences 6,82–89.

Grifﬁn, A. S.,Netto, K. &Peneaux, C. (2017). Neophilia, innovation and learning

in an urbanized world: a critical evaluation of mixed ﬁndings. Current Opinion in

Behavioral Sciences 16,15–22.

Grifﬁths, T. L. &Tenenbaum, J. B. (2007). From mere coincidences to meaningful

discoveries. Cognition 103(2), 180–226.

Gross, M. E.,Zedelius, C. M. &Schooler, J. W. (2020). Cultivating an

understanding of curiosity as a seed for creativity. Current Opinion in Behavioral

Sciences 35,77–82.

Gruber, M. J. &Fandakova, Y. (2021). Curiosity in childhood and adolescence—

what can we learn from the brain. Current Opinion in Behavioral Sciences 39, 178–184.

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.

Gweon, H.,Pelton, H.,Konopka, J. A. &Schulz,L.E.(2014). Sins of omission:

children selectively explore when teachers are under-informative. Cognition 132(3), 335–341.

Halina, M. (2015). There is no special problem of mindreading in nonhu man animals.

Philosophy of Science 82(3), 473–490.

Hall, B. A.,Melﬁ, V.,Burns, A.,Mcgill, D. M. &Doyle, R. E. (2018). Curious

creatures: a multi-taxa investigation of responses to novelty in a zoo environment.

PeerJ 6, e4454.

Hardy, J. H. III,Ness, A. M. &Mecca, J. (2017). Outside the box: epistemic

curiosity as a predictor of creative problem solving and creative performance.

Personality and Individual Differences 104, 230–237.

Harris, P. L. &Lane, J. D. (2014). Infants understand how testimony works. Topoi 33,

443–458.

Hausberger, M.,Henry, L.,Rethore, B.,Pougnault, L.,Kremers, D.,

Ro¨ ssler, C.,Aubry, C.,Cousillas, H.,Boye, M. &Lemasson, A. (2021).

When perceptual laterality vanishes with curiosity: a study in dolphins and

starlings. Laterality 26(1–2), 238–259.

Heinrich, B. (1995). Neophilia and exploration in juvenile common ravens, Corvus

corax.Animal Behaviour 50(3), 695–704.

Hembacher, E.,Demayo, B. &Frank, M. C. (2020). Children’s social information

seeking is sensitive to referential ambiguity. Child Development 91(6), e1178–e1193.

Herrmann, E.,Hare, B.,Cissewski, J. &Tomasello, M. (2011). A comparison of

temperament in nonhuman apes and human infants. Developmental Science 14(6),

1393–1405.

Horwitz, J. C. (2000). Mesolimbocortical and nigrostriatal dopamine responses to

salient non-reward events. Neuroscience 96(4), 651–656.

Huber, L.,Rechberger, S. &Taborsky, M. (2001). Social learning affects object

exploration and manipulation in keas, Nestor notabilis.Animal Behaviour 62(5), 945–954.

Inglis, I. R.,Forkman, B. &Lazarus, J. (1997). Free food or earned food? A review

and fuzzy model of contrafreeloading. Animal Behaviour 53(6), 1171–1191.

Inzani, E. L.,Kelley, L. A. &Boogert, N. J. (2022). Object neophilia in wild

herring gulls in urban and rural locations. Journal of Avian Biology 1–2, e03028.

Jaeggi, A. V.,Dunkel, L. P.,Van Noordwijk, M. A.,Wich, S. A.,Sura, A. A. &

van Schaik, C. P. (2010). Social learning of diet and foraging skills by wild

immature Bornean orangutans: implications for culture. American Journal of

Primatology 72(1), 62–71.

Jaeggi, A. V.,van Noordwijk, M. A. &van Schaik, C. P. (2008). Begging for

information: mother–offspring food sharing among wild Bornean orangutans.

American Journal of Primatology 70(6), 533–541.

James, W. (1890). The principles of Psychology Volume II.

Jepma, M.,Verdonschot, R. G.,Van Steenbergen, H.,Rombouts, S. A. &

Nieuwenhuis, S. (2012). Neural mechanisms underlying the induction and relief

of perceptual curiosity. Frontiers in Behavioral Neuroscience 6,5.

Kano, F. &Tomonaga, M. (2009). How chimpanzees look at pictures: a comparative

eye-tracking study. Proceedings of the Royal Society B: Biological Sciences 276(1664),

1949–1955.

Karl, S.,Boch, M.,Zamansky, A.,van der Linden, D.,Wagner, I. C.,

Vo¨ lter, C. J.,Lamm, C. &Huber, L. (2020). Exploring the dog–human

relationship by combining fMRI, eye-tracking and behavioural measures. Scientiﬁc

Reports 10(1), 22273.

Kaulfuß, P. &Mills, D. S. (2008). Neophilia in domestic dogs (Canis familiaris) and

its implication for studies of dog cognition. Animal Cognition 11, 553–556.

Kelleher, S. R .,Silla, A. J. &Byrne, P. G. (2018). Animalpersonality and behavioral

syndromes in amphibians: a review of the evidence, experimental approaches, and

implications for conservation. Behavioral Ecology and Sociobiology 72(5), 1–26.

Kelley, A. E.,Schochef, T. &Landry, C. F. (2004). Risk taking and novelty

seeking in adolescence: introduction to part I. Annals of the New York Academy of

Sciences 1021(1), 27–32.

Kelly, K. T. (2004). Justiﬁcation as truth-ﬁnding efﬁciency: how Ockham’s razor

works. Minds and Machines 14, 485–505.

Kendal, R. L.,Coe, R. L. &Laland, K. N. (2005). Age differences in neophilia,

exploration, and innovation in family groups of callitrichid monkeys. American

Journal of Primatology 66(2), 167–188.

Kidd, C. &Hayden, B. Y. (2015). The psychology and neuroscience of curiosity.

Neuron 88(3), 449–460.

Kidd, C.,Piantadosi, S. T. &Aslin, R. N. (2012). The goldilocks effect: human

infants allocate attention to visual sequences that are neither too simple nor too

complex. PLoS One 7, e36399.

Kinzler, K. D.,Dupoux, E. &Spelke, E. S. (2007). The native language of social

cognition. Proceedings of the National Academy of Sciences 104(30), 12577–12580.

KITA, S. (ed.) (2003). Pointing: Where Language, Culture, and Cognition Meet. Psychology

Press, New Jersey.

Kovacs, A. M.,Tauzin, T.,Teglas, E.,Gergely, G. &Csibra, G. (2014). Pointing as

epistemic request: 12-month-olds point to receive new information. Infancy 19(6), 543–557.

Kulkarni, T. D.,Narasimhan, K.,Saeedi, A. &Tenenbaum, J. (2016).

Hierarchical deep reinforcement learning: integrating temporal abstraction and

intrinsic motivation. Advances in Neural Information Processing Systems 29, 3675–3683.

Kummer, H. &Goodall, J. (1985). Conditions of innovative behaviour in primates.

Philosophical Transactions of the Royal Society of London. B, Biological Sciences 308(1135),

203–214.

Lachman, S. J. (1997). Learning is a process: toward an improved deﬁnition of

learning. The Journal of Psychology 131(5), 477–480.

Langston, R. F.,Stevenson, C. H.,Wilson, C. L.,Saunders, I. &Wood, E. R.

(2010). The role of hippocampal subregions in memory for stimulus associations.

Behavioural Brain Research 215(2), 275–291.

Laurent, P. A. (2008). The emergence of saliency and novelty responses from

reinforcement learning principles. Neural Networks 21(10), 1493–1499.

Lee, N.,Lazaro, V.,Wang, J. J.,Sen, H. H. &Lucca, K. (2023). Exploring

individual differences in infants’looking preferences for impossible events: the

early multidimensional curiosity scale. Frontiers in Psychology 13, 1015649.

Leslie, A. M.,Friedman, O. &German, T. P. (2004). Core mechanisms in ‘theory

of mind’.Trends in Cognitive Sciences 8(12), 528–533.

Lesort, T.,Lomonaco, V.,Stoian, A.,Maltoni, D.,Filliat, D. &Diaz-

Rodriguez, N. (2020). Continual learning for robotics: deﬁnition, framework,

learning strategies, opportunities and challenges. Information Fusion 58,52–68.

18 Soﬁa Forss and others

Levin, M. (2019). The computational boundary of a ‘self’: developmental bioelectricity

drives multicellularity and scale-free cognition. Frontiers in Psychology 10,2688.

Lewis, L. S. &Krupenye, C. (2022). Eye-tracking as a window into primate social

cognition. American Journal of Primatology 84(10), e23393.

Loewenstein, G. (1994). The psychology of curiosity: a review and reinterpretation.

Psychological Bulletin 116(1), 75–98.

Lucca, K. (2020). The development of information-requesting gestures in infancy and

their role in shaping learning outcomes. In The Questioning Child: Insights from Psychology

and Education (eds L. P. BUTLER,S.RONFARD and K. H. CORRIVEAU), pp. 89–117.

Cambridge University Press, Cambridge, UK.

Lucca, K. &Wilbourn, M. P. (2018). Communicating to learn: Infants’pointing

gestures result in optimal learning. Child Development 89(3), 941–960.

Lucca, K. &Wilbourn, M. P. (2019). The what and the how: information-seeking

pointing gestures facilitate learning labels and functions. Journal of Experi mental Child

Psychology 178, 417–436.

Lungarella, M.,Metta, G.,Pfeifer, R. &Sandini, G. (2003). Developmental

robotics: a survey. Connection Science 15, 151–190.

Luo, Y. &Baillargeon, R. (2005). Can a self-propelled box have a goal?

Psychological reasoning in 5-month-old infants. Psychological Science 16(8), 601–608.

Lyons, K. E. &Ghetti, S. (2011). The development of uncertainty monitoring in

early childhood. Child Development 82(6), 1778–1787.

Markey, A. &Loewenstein, G. (2014). Curiosity. In International Handbook of Emotions

in Education (eds R. PEKRUN and L. LINNENBRINK-GARCIA), pp. 238–255. Routledge,

New York.

Martins, C. I.,Schaedelin, F. C.,Mann, M.,Blum, C.,Mandl, I.,Urban, D.,

Grill, J.,Scho¨ sswender, J. &Wagner, R. H. (2012). Exploring novelty: a

component trait of behavioural syndromes in a colonial ﬁsh. Behaviour 149(2),

215–231.

Marvin, C. B.,Tedeschi, E. &Shohamy, D. (2020). Curiosity as the impulse to

know: common behavioral and neural mechanisms underlying curiosity and

impulsivity. Current Opinion in Behavioral Sciences 35,92–98.

Massen, J. J.,Antonides, A.,Arnold, A. M. K.,Bionda, T. &Koski, S. E. (2013).

A behavioral view on chimpanzee personality: exploration tendency, persistence,

boldness, and tool-orientation measured with group experiments. American Journal

of Primatology 75(9), 947–958.

McCrink, K. &Wynn, K. (2004). Large-number addition and subtraction by

9-month-old infants. Psychological Science 15(11), 776–781.

Mettke-Hofmann, C. (2014). Cognitive ecology: ecological factors, life-styles, and

cognition. Wiley Interdisciplinary Reviews: Cognitive Science 5(3), 345–360.

Mettke-Hofmann, C.,Rowe, K. C.,Hayden, T. J. &Canonine, V. (2006). Effects

of experience and object complexity on exploration in garden warblers (Sylvia borin).

Journal of Zoology 268(4), 405–413.

Mikhalevich, I.,Powell, R. &Logan, C. (2017). Is behavioural ﬂexibility evidence

of cognitive complexity? How evolution can inform comparative cognition. Interface

Focus 7(3), 20160121.

Miller, R.,Lambert, M. L.,Frohnwieser, A.,Brecht, K. F.,Bugnyar, T.,

Crampton, I.,Garcia-Pelegrin, E.,Gould, K.,Greggor, A. L.,

Izawa, E.-I.,Kelly, D. M.,Zhongqiu, L.,Yunchao, L.,Linh, B. L.,

Massen, J. J.,ET AL. (2022). Socio-ecological correlates of neophobia in corvids.

Current Biology 32(1), 74–85.

Mirolli, M. &Baldassarre, G. (2013). Functions and mechanisms of intrinsic

motivations. In Intrinsically Motivated Learning in Natural and Artiﬁcial Systems, pp. 49–

72. Springer Berlin, Heidelberg.

Moretti, L.,Hentrup, M.,Kotrschal, K. &Range, F. (2015). The inﬂuence of

relationships on neophobia and exploration in wolves and dogs. Animal Behaviour

107, 159–173.

Morgan, C. L. (1903). An introduction to comparative psychology, New ed., rev.

Nakano, T.,Watananbe, H.,Homae, F. &Taga, G. (2009). Prefrontal cortical

involvement in young infants’analysis of novelty. Cerebral Cortex 19(2), 455–463.

Nissen, H. W. (1931). A study of exploratory behavior in the white rat by means of

the obstruction method. In Animal Motivation Experimental Studies on the Albino Rat,

pp. 351–368. Columbia University Press, New York.

Noordewier, M. K. &van Dijk, E. (2017). Curiosity and time: from not knowing to

almost knowing. Cognition and Emotion 31(3), 411–421.

NYBORG, H. (ed.) (2003). The Scientiﬁc Study of General Intelligence: Tribute to Arthur Jensen.

Elsevier Science, Oxford, UK.

O’Hara, M.,Mioduszewska, B.,von Bayern, A.,Auersperg, A.,Bugnyar, T.,

Wilkinson, A.,Huber, L. &Gajdon, G. K. (2017). The temporal dependence of

exploration on neotic style in birds. Scientiﬁc Reports 7(1), 4742.

Oudeyer, P. Y. &Kaplan, F. (2007). What is intrinsic motivation? A typology of

computational approaches. Frontiers in Neurorobotics 1,6.

Oudeyer, P. Y.,Kaplan, F. &Hafner, V. V. (2007). Intrinsic motivation systems for

autonomous mental development. IEEE Transactions on Evolutionary Computation 11,

265–286.

Oudeyer, P. Y. &Smith, L. B. (2016). How evolution may work through curiosity-

driven developmental process. Topics in Cognitive Science 8(2), 492–502.

Parisi, G. I.,Tani, J.,Weber, C. &Wermter, S. (2018). Lifelong learning of

spatiotemporal representations with dual-memory recurrent self-organization.

Frontiers in Neurorobotics 12, 78.

Pathak, D.,Agrawal, P.,Efros, A. A. &Darrell, T. (2017). Curiosity-driven

exploration by self-supervised prediction. In International Conference on Machine

Learning, pp. 2778–2787. PMLR, Sydney.

PELLEGRINI,A.D.&SMITH, P. K. (eds) (2005). The Nature of Play: Great Apes and Humans.

Guilford Press, New York.

Perez, J. &Feigenson, L. (2021). Stable individual differences in infant s’responses to

violations of intuitive physics. Proceedings of the National Academy of Sciences 118(27),

e2103805118.

Perez, J. &Feigenson, L. (2022). Violations of expectation trigger infants to search

for explanations. Cognition 218, 104942.

Pfeifer, R. &Iida, F. (2004). Embodied artiﬁcial intelligence: trends and challenges.

Embodied Artiﬁcial Intelligence 3139,1–26.

Piantadosi, S. T.,Jara-Ettinger, J. &Gibson, E. (2014). Children’s learning of

number words in an indigenous farming-foraging group. Developmental Science 17(4),

553–563.

Pisula, W. (2020). Curiosity and Information Seeking in Animal and Human Behavior: A Review

the Literature and Data in Comparative Psychology, Animal Cognition, Ethology, Ontogenesis, and

Elements of Cognitive Neuroscience as they Relate to Animal Inquisitiveness. Universal

Publishers, Irvine, CA.

Pisula, W.,Turlejski, K. &Charles, E. P. (2013). Comparative psychology as

uniﬁed psychology: the case of curiosity and other novelty-related behavior. Review

of General Psychology 17(2), 224–229.

Pisula, W.,Turlejski, K.,Stryjek, R.,Nalecz-Tolak, A.,Grabiec, M. &

Djavadian, R. L. (2012). Response to novelty in the laboratory Wistar rat, wild-

captive WWCPS rat, and the gray short-tailed opossum (Monodelphis domestica).

Behavioural Processes 91(2), 145–151.

Powell, S. B.,Geyer, M. A.,Gallagher, D. &Paulus, M. P. (2004). The balance

between approach and avoidance behaviors in a novel object exploration paradigm

in mice. Behavioural Brain Research 152(2), 341–349.

Quinn, P. C.,Yahr, J.,Kuhn, A.,Slater, A. M. &Pascalis, O. (2002).

Representation of the gender of human faces by infants: a preference for female.

Perception 31(9), 1109–1121.

Reader, S. M. (2015). Causes of individual differences in animal exploration and

search. Topics in Cognitive Science 7(3), 451–468.

Redolat, R.,Perez-Martinez, A.,Carrasco, M. C. &Mesa, P. (2009).

Individual differences in novelty-seeking and behavioral responses to nicotine: a

review of animal studies. Current Drug Abuse Reviews 2(3), 230–242.

Reynolds, G. D. (2015). Infant visual attention and object recognition. Behavioural

Brain Research 285,34–43.

Rojas-Ferrer, I.,Thompson, M. J. &Morand-Ferron, J. (2020). Is exploration a

metric for information gathering? Attraction to novelty and plasticity in black-

capped chickadees. Ethology 126(4), 383–392.

Ronfard, S.,Zambrana, I. M.,Hermansen, T. K. &Kelemen, D. (2018).

Question-asking in childhood: a review of the literature and a framework for

understanding its development. Developmental Review 49, 101–120.

Ro¨ ssler, T.,Mioduszewska, B.,O’Hara, M.,Huber, L.,

Prawiradilag, D. M. &Auersperg, A. M. (2020). Using an innovation arena

to compare wild-caught and laboratory Gofﬁn’s cockatoos. Scientiﬁc Reports 10(1),

8681.

Roth, T. C.,Ladage, L. D. &Pravosudov, V. V. (2010). Learning capabilities

enhanced in harsh environments: a common garden approach. Proceedings of the

Royal Society B: Biological Sciences 277(1697), 3187–3193.

Russell, J. L.,Lyn, H.,Schaeffer, J. A. &Hopkins, W. D. (2011). The role of

socio-communicative rearing environments in the development of social and

physical cognition in apes. Developmental Science 14(6), 1459–1470.

Santucci, V. G.,Baldassarre, G. &Cartoni, E. (2019). Autonomous

reinforcement learning of multiple interrelated tasks. In Joint IEEE 9th International

Conference on Development and Learning and Epigenetic Robotics (ICDL-EpiRob), pp. 221–227.

IEEE, Oslo.

Santucci, V. G.,Baldassarre, G. &Mirolli, M. (2014). Cumulative learning

through intrinsic reinforcements. In Evolution, Complexity and Artiﬁcial Life, pp. 107–

122. Springer Berlin, Heidelberg.

Santucci, V. G.,Oudeyer, P. Y.,Barto, A. &Baldassarre, G. (2020). Editorial:

intrinsically motivated open-ended learning in autonomous robots. Frontiers in

Neurorobotics 13, 115.

Schillaci, G.,Ciria, A. &Lara, B. (2020). Tracking emotions: intrinsic motivation

grounded on multilevel prediction error dynamics. In 2020 Joint IEEE 10th

International Conference on Development and Learning and Epigenetic Robotics (ICDL-EpiRob).

IEEE, Valparaiso, Chile.

Schmidhuber, J. (1991a). A possibility for implementing curiosity and boredom in

model-building neural controllers. In Proceedings of the International Conference on

Simulation of Adaptive Behavior: From Animals to Animats, pp. 222–227. MIT

Press/Bradford Books, Cambridge, MA.

Transdisciplinary curiosity 19

Schmidhuber, J. (1991b). Curious model-building control systems. In Proceedings of the

International Joint Conference on Neural Networks, pp. 1458–1463. IEEE, Singapore.

Schultz, L. E. &Bonawitz, E. B. (2007). Serious fun: preschoolers engage in more

exploratory play when evidence is confounded. Developmental Psychology 43(4),

1045–1050.

Schultz, W. (2016). Dopamine reward prediction-error signalling: a two-component

response. Nature Reviews Neuroscience 17(3), 183–195.

Schuppli, C.,Forss, S. I. F.,Meulman, E. J.,Zweifel, N.,Lee, K. C.,

Rukumana, E.,Vogel, E. R.,van Noordwijk, M. A. &van Schaik, C. P.

(2016). Development of foraging skills in two orangutan populations: needing to

learn or needing to grow? Frontiers in Zoology 13,1–17.

Schwartz, C. E.,Kunwar, P. S.,Grev, D. N.,Moran, L. R.,Viner, J. C.,

Covino, J. M.,Kagan, J.,Stewart, E.,Snidman, N. C.,Vangel, M. G. &

Wallace, S. R. (2010). Structural differences in adult orbital and ventromedial

prefrontal cortex predicted by infant temperament at 4 months of age. Archives of

General Psychiatry 67(1), 78–84.

Shettleworth, S. J. (2010). Clever animals and killjoy explanations in comparative

psychology. Trends in Cognitive Sciences 14(11), 477–481.

Shettleworth, S. J. (2012). Darwin, Tinbergen, and the evolution of comparative

cognition. In The Oxford Handbook of Comparative Evolutionary Psychology, pp. 529–546.

Elsevier Publishing, Amsterdam.

Shin, D. D. &Kim, S. I. (2019). Homo curious: curious or interested? Educational

Psychology Review 31, 853–874.

Silvia, P. J. (2008). Interest—the curious emotion. Current Directions in Psychological

Science 17(1), 57–60.

Sim, Z. L. &Xu, F. (2017). Infants preferentially approach and explore the

unexpected. British Journal of Developmental Psychology 35(4), 596–608.

Sim, Z. L. &Xu, F. (2019). Another look at looking time: surprise as rational statistic al

inference. Topics in Cognitive Science 11(1), 154–163.

Siviter, H.,Charles Deeming, D.,Rosenberger, J.,Burman, O. H.,

Moszuti, S. A. &Wilkinson, A. (2017). The impact of egg incubation

temperature on the personality of oviparous reptiles. Animal Cognition 20(1), 109–116.

Sober, E. (2015). Is the scientiﬁc method a myth? Perspectives from the history and

philosophy of science. METODE Science Studies Journal 5, 195–199.

Stahl, A. E. &Feigenson, L. (2015). Observing the unexpected enhances infants’

learning and exploration. Science 348,91–94.

Sutton, R. S. (1991). Reinforcement learning architectures for animats. In Proceedings

of the First International Conference on Simulation of Adaptive Behavior on from Animals to

Animats, pp. 288–296. MIT Press, Cambridge, MA.

Sutton, R. S. &Barto, A. G. (2018). Reinforcement Learning: An Introd uction. MIT Press,

Cambridge, MA.

Szabo, B. &Ringler, E. (2023). Fear of the new? Geckos hesitate to attack novel prey,

feed near objects and enter a novel space. Animal Cognition 26(2), 537–549.

Takola, E.,Krause, E. T.,Mu

¨ller, C. &Schielzeth, H. (2021). Novelty at

second glance: a critical appraisal of the novel object paradigm based on meta-

analysis. Animal Behaviour 180, 123–142.

Taylor, M.,Esbensen, B. M. &Bennett, R. T. (1994). Children’s understanding of

knowledge acquisition: the tendency for children to report that they have always

known what they have just learned. Child Development 65(6), 1581–1604.

Teglas, E.,Girotto, V.,Gonzales, M. &Bonatti, L. L. (2007). Intuitions of

probabilities shape expectations about the future at 12 months and beyond.

Proceedings of the National Academy of Sciences 104(48), 19156–19159.

Thorndike, E. (1911). Animal intelligence: Experimentalstudies. Macmillan Press, New York.

Tian, X.,Silva, A. C. &Liu, C. (2021). The brain circuits and dynamics of curiosity -

driven behavior in naturally curious marmosets. Cerebral Cortex 31(9), 4220–4232.

Tinbergen, N. (1963). On aims and methods of ethology. Zeitschrift für tierpsyc hologie 20,

410–433.

Tomasello, M. &Call, J. (2004). The role of humans in the cognitive development

of apes revisited. Animal Cognition 7, 213–215.

Tomasello, M.,Carpenter, M. &Liszkowski, U. (2007). A new look at infant

pointing. Child Development 78(3), 705–722.

Turati, C.,Simion, F.,Milani, I. &Umilta, C. (2002). Newborns’preference for

faces: what is crucial? Developmental Psychology 38(6), 875–882.

Twomey, K. E. &Westermann, G. (2018). Curiosity-based learning in infants: a

neurocomputational approach. Developmental Science 21(4), e12629.

Valenza, E.,Simion, F.,Cassia,V.M.&Umilta, C. (1996). Face preference at

birth. Journal of Experimental Psychology: Human Perception and Performance 22(4),

892–903.

Valenzo, D.,Ciria, A.,Schillaci, G. &Lara, B. (2022). Grounding context in

embodied cognitive robotics. Frontiers in Neurorobotics 16, 843108.

van de Waal, E. &Bshary, R. (2011). Contact with human facilities appears to

enhance technical skills in wild vervet monkeys (Chlorocebus aethiops). Folia

Primatologica 81(5), 282–291.

van Schaik, C. P.,Burkart, J.,Damerius, L.,Forss, S. I.,Koops, K.,van

Noordwijk, M. A. &Schuppli, C. (2016). The reluctant innovator: orangutans

and the phylogeny of creativity. Philosophical Transactions of the Royal Society B:

Biological Sciences 371(1690), 20150183.

Wang, M. Z. &Hayden, B. Y. (2019). Monkeys are curious about counterfactual

outcomes. Cognition 189,1–10.

Wang, M. Z. &Hayden, B. Y. (2021). Latent learning, cognitive maps, and curiosity.

Current Opinion in Behavioral Sciences 38,1–7.

Wang, S. H.,Baillargeon, R. &Brueckner, L. (2004). Young infants’reasoning

about hidden objects: evidence from violation-of-expectation tasks with test trials

only. Cognition 93(3), 167–198.

Wang, S. H.,Zhang, Y. &Baillargeon, R. (2016). Young infants view physically

possible support events as unexpected: new evidence for rule learning. Cognition

157, 100–105.

Warner, M. (2007). Sources and methods for the study of intelligence. In Handbook of

intelligence studies, pp. 35–45. Routledge Press, New York.

Washburne, J. N. (1936). The deﬁnition of learning. Journal of Educational Psychology

27(8), 603–611.

Wendler, D. (2014). Should protections for research with humans who cannot

consent apply to research with nonhuman primates? Theoretical Medicine and Bioethics

35, 157–173.

Werchan, D. M.,Collins, A. G.,Frank, M. J. &Amso, D. (2016). Role of

prefrontal cortex in learning and generalizing hierarchical rules in 8-month-old

infants. Journal of Neuroscience 36(40), 10314–10322.

White, R. W. (1959). Motivation reconsidered: the concept of competence.

Psychological Review 66, 297–333.

Wittmann, B. C.,Bunzeck, N.,Dolan, R. J. &Du

¨zel, E. (2007). Anticipation of

novelty recruits reward system and hippocampus while promoting recollection.

NeuroImage 38(1), 194–202.

Wu, Y. &Gweon, H. (2019). Surprisingly unsurprising! Infants’looking time at

probable vs. improbable events is modulated by others’expressions of surprise.

https://doi.org/10.31234/osf.io/8whuv.

Wynn, K. (1992). Children’s acquisition of the number words and the counting system.

Cognitive Psychology 24(2), 220–251.

Wynne, C. D. (2004). The perils of anthropomorphism. Nature 428(6983), 606.

Xiong, F.,Liu, Z.,Huan, K.,Yang, X. &Qiao, H. (2021). State primitive learning

to overcome catastrophic forgetting in robotics. Cognitive Computation 13, 394–402.

Zettersten, M. &Saffran, J. R. (2021). Sampling to learn words: adults and

children sample words that reduce referential ambiguity. Developmental Science 24(3),

e13064.

Zhai, X.,Puigcercer, J.,Kolesnikov, A.,Ruyssen, P.,Riquelme, C.,

Lucic, M.,Djolonga, J.,Susano Pinto, A.,Neumann, M.,

Dosovitskiy, A.,Beyer, L.,Bachem, O.,Tschannen, M.,Michalski, M.,

Bousquet, O.,ET AL. (2019). A large-scale study of representation learning with

the visual task adaptation benchmark. arXiv preprint arXiv:1910.04867.

(Received 19 May 2023; revised 8 January 2024; accepted 12 January 2024 )

20 Soﬁa Forss and others

The curious interpretation of novel object recognition tests

Article

Mar 2025
TRENDS NEUROSCI

The role of exploration and exploitation in primate communication

Article

Full-text available

Jan 2025
Proc Biol Sci

The concepts of social learning and exploration have been central to debates in comparative cognition research. While their roles in the origins of human cumulative culture on the one hand and creativity on the other have been highlighted, the two concepts have mostly been studied separately. In this article, we examine the relationship between adopting similar or different behaviours within a group, focusing on how exploration and exploitation shape primate communication systems. Using a comparative approach, we discuss how similarity and differentiation of communicative behaviour can be viewed as two endpoints on a continuum, impacting both individual- and group-level behavioural variation. While group-level variation is evident in some ape behaviours (e.g. foraging traditions), individual variation in communicative behaviour appears to outweigh group-level differences, making a widespread communicative culture in apes unlikely. Drawing parallels to language acquisition in human infants, we propose that ape communication follows an exploration–exploitation trajectory, with initial exploration gradually giving way to focused exploitation of genetically predisposed and/or individually developed communicative repertoires. By integrating the individual and social learning processes underlying communicative behaviour, we can gain a deeper understanding of how exploration–exploitation tensions shape communication systems across species.

The State of Global Catastrophic Risk Research: A Bibliometric Review

Preprint

Full-text available

Nov 2024

The global catastrophic risk (GCR) and existential risk (ER) literature focuses on analysing and preventing potential major global catastrophes including a human extinction event. Over the past two decades, the field of GCR/ER research has grown considerably. However, there has been little meta-research on the field itself. How large has this body of literature become? What topics does it cover? Which fields does it interact with? What challenges does it face? To answer these questions, here we present the first systematic bibliometric analysis of the GCR/ER literature. We consider all 3,437 documents in the OpenAlex database that mention either GCR or ER, and use bibliographic coupling (two documents are considered similar when they share many references) to identify ten distinct emergent research clusters in the GCR/ER literature. These clusters 2 align in part with commonly identified drivers of GCR, such as advanced artificial intelligence (AI), climate change, and pandemics, or discuss the conceptual foundations of the GCR/ER field. However, the field is much broader than these topics, touching on disciplines as diverse as economics, climate modeling, agriculture, psychology, and philosophy. The metadata reveal that there are around 150 documents published on GCR/ER each year, the field has highly unequal gender representation, most research is done in the US and the UK, and many of the published articles come from a small subset of authors. We recommend creating new conferences and potentially new journals where GCR/ER focused research can aggregate, making gender and geographic diversity a higher priority, and fostering synergies across clusters to think about GCR/ER in a more holistic way. We also recommend building more connections to new fields and neighboring disciplines, such as systemic risk and policy, to encourage cross-fertilisation and the broader adoption of GCR/ER research.

To know or not to know? Curiosity and the value of prospective information in animals

Article

Full-text available

Oct 2024

Humans and other animals often seek instrumental information to strategically improve their decisions in the present. Our curiosity also leads us to acquire non-instrumental information that is not immediately useful but can be encoded in memory and stored for use in the future by means of episodic recall. Despite its adaptive benefits and central role in human cognition, questions remain about the cognitive mechanisms and evolutionary origins that underpin curiosity. Here, we comparatively review recent empirical studies that some authors have suggested reflects curiosity in nonhuman animals. We focus on findings from laboratory tasks in which individuals can choose to gain advanced information about uncertain future outcomes, even though the information cannot be used to increase future rewards and is often costly. We explore the prevalence of preferences in these tasks across animals, discuss the theoretical advances that they have promoted, and outline some limitations in contemporary research. We also discuss several features of human curiosity that can guide future empirical research aimed at characterising and understanding curiosity in animals. Though the prevalence of curiosity in animals is actively debated, we surmise that investigating behavioural candidates for curiosity-motivated behaviour in a broader range of species and contexts, should help promote theoretical advances in our understanding of cognitive principles and evolutionary pressures that support curiosity-driven behaviour.

Early Emergence of Metacognition in Rhesus Monkeys

Article

Mar 2025
DEVELOPMENTAL SCI

Metacognition, or monitoring and controlling one's knowledge, is a key feature of human cognition. Accumulating evidence shows that foundational forms of metacognition are already present in young infants and then scaffold later‐emerging skills. Although many animals exhibit cognitive processes relevant to metacognition, it is unclear if other species share the developmental trajectories seen in humans. Here, we examine the emergence of metacognitive information‐seeking in rhesus monkeys ( Macaca mulatta ). We presented a large sample of semi‐free‐ranging monkeys, ranging from juvenility to adulthood, with a one‐shot task where they could seek information about a food reward by bending down to peer into a center vantage point in an array of tubes. In the hidden condition, information‐seeking was necessary as no food was visible on the apparatus, whereas in the visible control, condition information‐seeking was not necessary to detect the location of the reward. Monkeys sought information at the center vantage point more often when it was necessary than in the control condition, and younger monkeys already showed competency similar to adults. We also tracked additional monkeys who voluntarily chose not to approach to assess monkeys’ ability to actively infer opportunities for information‐seeking, and again found similar performance in juveniles and adults. Finally, we found that monkeys were overall slower to make metacognitive inferences than to approach known reward, and that younger monkeys were specifically slower to detect opportunities for information‐seeking compared to adults. These results indicate that many features of mature metacognition are already detectable in young monkeys, paralleling evidence for “core metacognition” in infant humans.

Unveiling the Enigma of Blind Box Impulse Buying Curiosity: The Moderating Role of Price Consciousness

Article

Full-text available

Dec 2024

This study explores the causes of curiosity-driven impulse buying in blind boxes using the Stimulus-Organism-Response (S-O-R) model and adaptation level theory. It examines how store environment and product factors contribute to customer curiosity, incorporating price consciousness into the overall framework. Insights from an online survey of 306 Chinese respondents indicate that environmental factors and specific product characteristics positively influence consumer curiosity, while price consciousness has a negative moderate effect. The findings also show that customer curiosity stimulates impulse buying behavior and mediates the relationship between store atmosphere, surprise, and perceived novelty. This study identifies both objective and subjective factors behind blind box impulse buying and offers relevant suggestions for governments and consumers on managing impulse buying.

Flexible information-seeking in chimpanzees

Article

Jul 2024

The Role of Umwelt in Animal Curiosity: A Within and Between Species Comparison of Novelty Exploration in Mongooses

Article

Full-text available

Nov 2023

In its broadest sense, curiosity has been described as an intrinsic motivation to acquire novel information; this ‘novelty-seeking’ is notably in the absence of any immediate reward. One way to examine information seeking in animals, has been to present animals with novel objects and measure the way animals gather information through exploration. While this is a standardized and common paradigm, few studies have focused on what factors influence how animals perceive novelty, whilst considering the predicted impacts of an animal’s ‘umwelt’ on exploration. In this study we assessed explorative behaviors in mongooses through both an intra and interspecific view. First, in meerkats (Suricata suricatta) tested in their natural environment, we established that they distinguish familiar from unfamiliar and show increased exploration of novel objects. We also found that odor influenced the meerkats’ explorative action, so that non-odorous items were manipulated longer. Presence of conspecifics influenced approaches to novelty, but not the exploration per se. Human presence interacted with an individual’s level of habituation to impact approaches and exploration of novelty and we found a strong captivity effect on exploration between captive and wild meerkats. Between species analysis showed that yellow mongooses (Cynictis penicillate), a less social mongoose than meerkats, showed higher levels of novelty exploration, when tested within the same habitat. Thus, these findings show that multiple factors, including perceptual abilities (merkwelt) and social factors (sozialwelt), are decisive for when and how animals explore their environment and must be considered both when designing novelty paradigm tests and their interpretations.

The endangered brain: actively preserving ex-situ animal behaviour and cognition will benefit in-situ conservation

Article

Full-text available

Aug 2023

Endangered species have small, unsustainable population sizes that are geographically or genetically restricted. Ex-situ conservation programmes are therefore faced with the challenge of breeding sufficiently sized, genetically diverse populations earmarked for reintroduction that have the behavioural skills to survive and breed in the wild. Yet, maintaining historically beneficial behaviours may be insufficient, as research continues to suggest that certain cognitive-behavioural skills and flexibility are necessary to cope with human-induced rapid environmental change (HIREC). This paper begins by reviewing interdisciplinary studies on the ‘captivity effect’ in laboratory, farmed, domesticated and feral vertebrates and finds that captivity imposes rapid yet often reversible changes to the brain, cognition and behaviour. However, research on this effect in ex-situ conservation sites is lacking. This paper reveals an apparent mismatch between ex-situ enrichment aims and the cognitive-behavioural skills possessed by animals currently coping with HIREC. After synthesizing literature across neuroscience, behavioural biology, comparative cognition and field conservation, it seems that ex-situ endangered species deemed for reintroduction may have better chances of coping with HIREC if their natural cognition and behavioural repertoires are actively preserved. Evaluating the effects of environmental challenges rather than captivity per se is recommended, in addition to using targeted cognitive enrichment.

Exploring individual differences in infants’ looking preferences for impossible events: The Early Multidimensional Curiosity Scale

Article

Full-text available

Feb 2023

Infants are drawn to events that violate their expectations about the world: they look longer at physically impossible events, such as when a car passes through a wall. Here, we examined whether individual differences in infants’ visual preferences for physically impossible events reflect an early form of curiosity, and asked whether caregivers’ behaviors, parenting styles, and everyday routines relate to these differences. In Study 1, we presented infants (N = 47, Mage = 16.83 months, range = 10.29–24.59 months) with events that violated physical principles and closely matched possible events. We measured infants’ everyday curiosity and related experiences (i.e., caregiver curiosity-promoting activities) through a newly developed curiosity scale, The Early Multidimensional Curiosity Scale (EMCS). Infants’ looking preferences for physically impossible events were positively associated with their score on the EMCS, but not their temperament, vocabulary, or caregiver trait curiosity. In Study 2A, we set out to better understand the relation between the EMCS and infants’ looking preferences for physically impossible events by assessing the underlying structure of the EMCS with a larger sample of children (N = 211, Mage = 47.63 months, range = 10.29–78.97 months). An exploratory factor analysis revealed that children’s curiosity was comprised four factors: Social Curiosity, Broad Exploration, Persistence, and Information-Seeking. Relatedly, caregiver curiosity-promoting activities were composed of five factors: Flexible Problem-Solving, Cognitive Stimulation, Diverse Daily Activities, Child-Directed Play, and Awe-Inducing Activities. In Study 2B (N = 42 infants from Study 1), we examined which aspects of infant curiosity and caregiver behavior predicted infants’ looking preferences using the factor structures of the EMCS. Findings revealed that infants’ looking preferences were uniquely related to infants’ Broad Exploration and caregivers’ Awe-Inducing Activities (e.g., nature walks with infants, museum outings). These exploratory findings indicate that infants’ visual preferences for physically impossible events may reflect an early form of curiosity, which is related to the curiosity-stimulating environments provided by caregivers. Moreover, this work offers a new comprehensive tool, the Early Multidimensional Curiosity Scale, that can be used to measure both curiosity and factors related to its development, starting in infancy and extending into childhood.

Object neophilia in wild herring gulls in urban and rural locations

Article

Full-text available

Dec 2022
J AVIAN BIOL

Living with increasing urbanisation and human populations requires resourcefulness and flexibility in wild animals' behaviour. Animals have to adapt to anthropogenic novelty in habitat structure and resources that may not resemble, or be as beneficial as, natural resources. Herring gulls Larus argentatus increasingly reside in towns and cities to breed and forage, yet how gulls are adjusting their behaviour to life in urban areas is not yet fully understood. This study investigated wild herring gulls' responses to novel and common anthropogenic objects in urban and rural locations. We also examined whether gulls' age influenced their object response behaviour. We found that, out of the 126 individual gulls presented with objects, 34% approached them. This suggests that the majority of targeted gulls were wary or lacked interest in the experimental set‐up. Of the 43 gulls that approached the objects, we found that those tested in urban locations approached more slowly than their rural counterparts. Overall, gulls showed no preference for either novel or common anthropogenic objects, and age did not influence likelihood of approach, approach speed or object choice. Individuals paid most attention to the object they approached first, potentially indicative of individual preferences. Our findings indicated that most herring gulls are not as attracted to anthropogenic objects as anecdotal reports have suggested. Covering up obvious food rewards may thus help mitigate human–gull conflict over anthropogenic food sources.

Fear of the new? Geckos hesitate to attack novel prey, feed near objects and enter a novel space

Article

Full-text available

Sep 2022
ANIM COGN

Neophobia, the fear of novelty, is an ecologically important response which enables animals to avoid potentially harmful situations. Neophobia is a cognitive process by which individuals distinguish novelty from familiarity. In this study, we aimed to quantify this cognitive process in captive tokay geckos (Gekko gecko) across three contexts: when encountering novel prey, foraging near novel objects and entering a novel space. We also investigated individual consistency across trials using different novel stimuli, and correlation of individual responses across the three contexts. We found that geckos hesitate to attack novel prey and prey close to objects (familiar and novel). Geckos hesitated the most when entering novel space. Repeatability of behaviour within and across contexts was low (R = 0.101-0.190) indicating that neophobia might not be expressed similarly across contexts. The strength of a neophobic response can indicate how anxious or curious an individual is. This test has great potential to help answer questions about how captivity, enrichment, rearing environment and cognition affect fear responses in different contexts in lizards. By studying reptiles, we can better understand the universality of what is known about the causes leading to difference in neophobia across individuals and species.

Grounding Context in Embodied Cognitive Robotics

Article

Full-text available

Jun 2022

Biological agents are context-dependent systems that exhibit behavioral flexibility. The internal and external information agents process, their actions, and emotions are all grounded in the context within which they are situated. However, in the field of cognitive robotics, the concept of context is far from being clear with most studies making little to no reference to it. The aim of this paper is to provide an interpretation of the notion of context and its core elements based on different studies in natural agents, and how these core contextual elements have been modeled in cognitive robotics, to introduce a new hypothesis about the interactions between these contextual elements. Here, global context is categorized as agent-related, environmental, and task-related context. The interaction of their core elements, allows agents to first select self-relevant tasks depending on their current needs, or for learning and mastering their environment through exploration. Second, to perform a task and continuously monitor its performance. Third, to abandon a task in case its execution is not going as expected. Here, the monitoring of prediction error, the difference between sensorimotor predictions and incoming sensory information, is at the core of behavioral flexibility during situated action cycles. Additionally, monitoring prediction error dynamics and its comparison with the expected reduction rate should indicate the agent its overall performance on executing the task. Sensitivity to performance evokes emotions that function as the driving element for autonomous behavior which, at the same time, depends on the processing of the interacting core elements. Taking all these into account, an interactionist model of contexts and their core elements is proposed. The model is embodied, affective, and situated, by means of the processing of the agent-related and environmental core contextual elements. Additionally, it is grounded in the processing of the task-related context and the associated situated action cycles during task execution. Finally, the model proposed here aims to guide how artificial agents should process the core contextual elements of the agent-related and environmental context to give rise to the task-related context, allowing agents to autonomously select a task, its planning, execution, and monitoring for behavioral flexibility.

Reinforcement Learning Architectures for Animats

Chapter

Feb 1991

Richard S. Sutton

These sixty contributions from researchers in ethology, ecology, cybernetics, artificial intelligence, robotics, and related fields delve into the behaviors and underlying mechanisms that allow animals and, potentially, robots to adapt and survive in uncertain environments. They focus in particular on simulation models in order to help characterize and compare various organizational principles or architectures capable of inducing adaptive behavior in real or artificial animals. Jean-Arcady Meyer is Director of Research at CNRS, Paris. Stewart W. Wilson is a Scientist at The Rowland Institute for Science, Cambridge, Massachusetts. Bradford Books imprint

A Possibility for Implementing Curiosity and Boredom in Model-Building Neural Controllers

Chapter

Feb 1991

Juergen Schmidhuber

Curiosity as a metacognitive feeling

Article

Feb 2023
COGNITION

Curious information-seeking is known to be a key driver for learning, but characterizing this important psychological phenomenon remains a challenge. In this article, we argue that solving this challenge requires qualifying the relationships between metacognition and curiosity. The idea that curiosity is a metacognitive competence has been resisted: researchers have assumed both that young children and non-human animals can be genuinely curious, and that metacognition requires conceptual and culturally situated resources that are unavailable to young children and non-human animals. Here, we argue that this resistance is unwarranted given accumulating evidence that metacognition can be deployed procedurally, and we defend the view that curiosity is a metacognitive feeling. Our metacognitive view singles out two monitoring steps as a triggering condition for curiosity: evaluating one's own informational needs, and predicting the likelihood that explorations of the proximate environment afford significant information gains. We review empirical evidence and computational models of curiosity, and show that they fit well with this metacognitive account, while on the contrary, they remain difficult to explain by a competing account according to which curiosity is a basic attitude of questioning. Finally, we propose a new way to construe the relationships between curiosity and the human-specific communicative practice of questioning, discuss the issue of how children may learn to express their curiosity through interactions with others, and conclude by briefly exploring the implications of our proposal for educational practices.

Learning with certainty in childhood

Article

Sep 2022
TRENDS COGN SCI

Learners use certainty to guide learning. They maintain existing beliefs when certain, but seek further information when they feel uninformed. Here, we review developmental evidence that this metacognitive strategy does not require reportable processing. Uncertainty prompts nonverbal human infants and nonhuman animals to engage in strategies like seeking help, searching for additional information, or opting out. Certainty directs children’s attention and active learning strategies and provides a common metric for comparing and integrating conflicting beliefs across people. We conclude that certainty is a continuous, domain-general signal of belief quality even early in life.

A transdisciplinary view on curiosity beyond linguistic humans: animals, infants, and artificial intelligence

Abstract and Figures

Recommended publications

The Role of Umwelt in Animal Curiosity: A Within and Between Species Comparison of Novelty Explorati...

Urbanization does not increase “object curiosity” in vervet monkeys, but semi-urban individuals sele...

Searching for Information, from Infancy to Adolescence

Information-Seeking Behaviors Show 3-Year-Old Children’s Metacognitive Abilities During Novel Word L...