Investigating intention in non-human animals. Part I. States of art and non-art

Abstract

Is the analytical framework used by ethologists sufficient to study the mental states of non-human animals (NHAs) at the appropriate level of complexity? To address this question our strategy was to i) reveal the experimental and analytic habits of scientists of different disciplines in the literature, and ii) use "intention" as a vector in an interdisciplinary prospect of the study of NHAs mental states. Our own intention was to outline the specific orientations and possible impasses of the ethological analytical framework which limits the consideration of NHAs intentions. We conducted a bibliometric analysis of the scientific literature published between 2016 and 2020 in two steps: 1. through a first corpus, we identified the terms used in studies of NHAs intentions and 2. on this basis, 111 articles related to intentions in NHAs were selected. By analysing them using a co-occurrences network of the authors’ keywords, ten scientific approaches to intention in NHAs were identified. Our main findings are that i) the term « intention » is very seldom used in studies of NHAs; ii) approaches developed in humans are rarely transposed in these studies; and iii) in such few studies, it is not the NHAs intentions which are under question, but the link between NHAs and human intentions. This study highlights the limitations of the current theoretical framework used to study non-human animals’ cognition, which does not allow for the full spectrum of non-human cognitive specificities.

Full text

Investigating intention in non-human animals. Part I.
States of art and non-art
Anne-Lise Dauphiné-Morer (  anne-lise.dauphine-morer@inrae.fr )
INRAE https://orcid.org/0000-0003-4049-6891
Franck Zenasni
LaPEA https://orcid.org/0000-0002-9340-0999
Alain Boissy
INRAE https://orcid.org/0000-0001-7688-4541
Muriel Mambrini-Doudet
INRAE https://orcid.org/0000-0002-8429-5664
Research Article
Keywords: Intention, Bibliometric analysis, Non-human animals, Science mapping, Cognition, Behaviour
Posted Date: December 11th, 2023
DOI: https://doi.org/10.21203/rs.3.rs-2941491/v2
License:   This work is licensed under a Creative Commons Attribution 4.0 International License. 
Read Full License
Additional Declarations: The authors declare no competing interests.

1
Investigating intention in non-human animals. 1
Part I. States of art and non-art 2
3
Authors 4
Anne-Lise Dauphiné-Morer1,2*, Franck Zenasni2, Alain Boissy1, Muriel Mambrini-Doudet3
5
1 INRAE, UMR1213 Herbivores, F-63122 Saint-Genès-Champanelle, France 6
2 Université Paris Cité, Univ Gustave Eiffel, LaPEA, F-92100 Boulogne-Billancourt, France 7
3 INRAE, CODIR - Collège de Direction, F-75007 Paris, France 8
* Corresponding author: anne-lise.dauphine-morer@cri-paris.org 9
10
ORCID: 11
AL. Dauphiné-Morer: 0000-0003-4049-6891 12
F. Zenasni: 0000-0002-9340-0999 13
A. Boissy: 0000-0001-7688-4541 14
M. Mambrini-Doudet: 0000-0002-8429-5664 15
Abstract 16
Is the analytical framework used by ethologists sufficient to study the mental states of non-human animals (NHAs) 17
at the appropriate level of complexity? To address this question our strategy was to i) reveal the experimental and 18
analytic habits of scientists of different disciplines in the literature, and ii) use "intention" as a vector in an 19
interdisciplinary prospect of the study of NHAs mental states. Our own intention was to outline the specific 20
orientations and possible impasses of the ethological analytical framework which limits the consideration of NHAs 21
intentions. We conducted a bibliometric analysis of the scientific literature published between 2016 and 2020 in 22
two steps: 1. through a first corpus, we identified the terms used in studies of NHAs intentions and 2. on this basis, 23
111 articles related to intentions in NHAs were selected. By analysing them using a co-occurrences network of the 24
authors’ keywords, ten scientific approaches to intention in NHAs were identified. Our main findings are that i) 25
the term « intention » is very seldom used in studies of NHAs; ii) approaches developed in humans are rarely 26
transposed in these studies; and iii) in such few studies, it is not the NHAs intentions which are under question, 27
but the link between NHAs and human intentions. This study highlights the limitations of the current theoretical 28
framework used to study non-human animals’ cognition, which does not allow for the full spectrum of non-human 29
cognitive specificities. 30
Keywords 31
Intention; Bibliometric analysis; non-human animals; Science mapping; Cognition; Behaviour 32
Statements and declarations 33
Conflict of interests The authors declare they have no conflict of interest to report. 34
Acknowledgments 35
The authors warmly thank the FAQ Platform CorText, Cécilia Guerra, Marie-Christine Meunier-Salaün, Marc 36
Santolini. The PhD of Anne-Lise Dauphiné-Morer is undertaken on the École Universitaire de Recherche 37
Interdisciplinaire de Paris (EURIP) graduate program benefiting from a national grant (ANR-17-EURE-0012) 38
supported by the Bettencourt-Schuller foundation. This PhD is funded by the French National Research Institute 39
for Agriculture, Food and the Environment (INRAE). This paper have an preprint DOI 40
https://doi.org/10.21203/rs.3.rs-2941491/v1]https://doi.org/10.21203/rs.3.rs-2941491/v1] 41

2
Introduction 42
43
Do ethologists have an adequate analytical framework at their disposal to consider the extent of the mental states 44
of non-human animals at the level of complexity at which it can now be studied? 45
The increase in knowledge about non-human animals’ mental states has clearly modified our perception of them. 46
Researchers and society have enriched their initial representation of the animal scheme by adding considerations 47
of sentience, conscious process (Le Neindre et al., 2018) and will (Greiveldinger et al., 2011; Heyes and Dickinson, 48
1990), among others. Each of these considerations is set against a different theoretical background, which means 49
that the foundations underlying the acceptance of mental states are very discrepant. Why is this the case? Does it 50
impede us from embracing greater complexity? 51
At the root of the development of cognitive sciences, for humans and non-human animals, we find Brentano’s 52
work. Brentano was a philosopher at the end of the 19th century who worked mainly on the question of mental 53
representations through the concept of intentionality. In his book “Psychology From an Empirical Standpoint” 54
(1874), he defined intentionality as the large set of mental states related to representations, the "aboutness". Since 55
this major input, the theoretical definition of the concept of intentionality and its experimental effects have been 56
discussed in the field of ethology. In the following, we propose a non-exhaustive overview of the theoretical and 57
experimental frameworks built on this concept. An overview of these frameworks can help to infer the potential 58
impact of such rooting. 59
The concept of intentionality, as developed by Brentano, claims to address the issue of mental representations 60
(Jacob, 2019). Based on this proposal, several philosophers have developed theoretical and practical frameworks 61
that support the experimental exploration of mental states so defined. One of the most important proposals has 62
been Dennett’s “Intentional Systems Theory” (1983), in which he organised intentionality into four grades (or 63
orders), representing different levels of complexity. To our knowledge, most work on intentionality uses such an 64
ordering system. In addition, Dennett (2009, 1983) has also developed a practical approach or, in other words, a 65
framework for experimenting with and studying intentionality in other species. In his article Intentional systems 66
in cognitive ethology: The "Panglossian paradigm" (1983), he proposes considering the animal under study as an 67
intentional system by describing its behaviour through "intentional idioms". These descriptions would, according 68
to Dennett, make it possible not only to explore the capacities of other species more widely, but also to identify 69
the levels of intentionality. For their part, Heyes and Dickinson (1990) proposed the Intentional theory of action, 70
which addresses the content of mental states in terms of beliefs, desires and practical inference processes. With 71
this approach, they claim to be able, through minimum analysis, to discern true intentional processes from non-72
intentional processes. Bratman's (1987) theory was also based on the tryptic belief-desire-intention (BDI), on 73
which a cognitive model of behaviour prediction is based (BDI model). Also in the context of the study of 74
intentionality, Dretske's (1988, 2003) made a significant effort to naturalise intentionality, i.e. to explain it by its 75
causality (for a critical view, see Proust, 1995, 1999). 76
Such flourishing theoretical work was particularly mobilised by scientists willing to explore and understand 77
mental states and cognition for individuals without (or not having mastered) human verbal language. For 78
ethologists, such theories were particularly useful for adapting these methods not based on verbal report to non-79
human animals (Boissy et al., 2007). These experimental approaches usually infer mental states and their 80
underlying cognitive processes (inner) from behavioural and physiological expressions (outer) – for a critical view 81
of the validity of this approach, see Dretske, 1980. In short, using the theoretical framework that places 82
intentionality as the expression of mental representations, ethologists explore intentionality through behaviours 83
and actions. Now, if we follow Brentano’s theory, intentions play a particular role in the etiology of action (Jacob, 84
2019). It follows that it will be practically difficult, if not impossible, to separate the study of intentionality from 85
that of intentions. 86
Based on this knowledge, assessing the mental states of non-human animals would involve, among other things, 87
assessing their intentions as well as their expressions of the same, i.e. being able to recognise and understand what 88
an intention and its expression might be for a given species. The properties (or attributes) of the intentions, to 89
connect the inner and outer, and make the actions representative of the mental states, as outlined by Brentano, are 90
used to develop the analytical framework of ethology without reidentifying that they are at the root. Specifically 91
studying intentions should thus re-evaluate and ultimately add to the current theoretical framework of ethology. 92

3
Since Brentano, the study of mental states has been drastically enriched, offering new investigative pathways for 93
studying non-human animals. 94
Several behaviours have been explored as the expression of intentions in other species. Among the behaviours 95
studied to access and/or evaluate intentions, goal-directed behaviours seem central (Vasconcelos et al., 2012). 96
These behaviours have been notably studied in relation to communication, particularly in great apes (Byrne et al., 97
2017; Gupta and Sinha, 2019; Leavens et al., 2005; Leavens and Hopkins, 1998; Molesti et al., 2020; Schel et al., 98
2013). Another important part of the study of the expression of intentions through communication is the study of 99
pointing behaviours (gestures to show someone something; for a detailed review, see Krause et al., 2018). 100
Intentions, or their expressions, therefore seem to be assessable through the capacity of a given individual to orient 101
their behaviour and/or attention towards a goal or a subject in a communication situation. 102
Intentions are also studied through the question of the individual’s mental state. For example, intention 103
movements are studies of the expression of motivation (Fischer and Zinner, 2011). Complex approaches linking 104
emotions, intentions and behavioural adaptability around the idea of goal desirability have also been developed 105
(Griffin and Speck, 2004; Seth et al., 2005). In an even more complex understanding, Dickinson and Balleine 106
(2000, p.202) proposed that “the capacity for goal-directed action requires not only the evolution of intentional 107
representations, but also the co-evolution of an interface between these representations and the animal’s biological 108
responses to the goal objects, events, or states”. In humans, the concept of intention is also closely associated with 109
an individual's expectations (Ajzen, 2011; Helfer et al., 2015; Kytö et al., 2019) and satisfactions (Diener et al., 110
2009). This approach has resulted in the widely used “Theory of Planned Behaviour” which claims to predict 111
individual behaviours (for a review, see Ajzen, 1991). In short, from common roots, a variety of theoretical 112
frameworks have developed and are now used to study intentions. 113
In addition to the approach to the inner state of the individual, studies on intentions are closely linked to those 114
of social behaviour; in particular, through the capacity of an individual to perceive the intention of another. 115
Baldwin and Baird (2001), for instance, propose that the relationship with others relies heavily on judgments 116
concerning the underlying intention of a given behaviour. In other words, it is claimed that we are not interested 117
in the behaviour of others for its own sake, but for what it reveals in terms of intentions. This aspect has been 118
widely studied through the “Theory of mind” (for a review in great apes, see Towner, 2010). Other studies explore 119
intentions as an interface between non-human animals and humans, through the recognition of others’ intentions 120
(for examples: in horses, see Trösch et al., 2020; in dogs, see Völter et al., 2023), in the species through the concept 121
of “shared intentionality” (for a definition of shared intentionality and related behaviours, see Tomasello and 122
Carpenter, 2007, and for an example of study, see Genty et al., 2020). In short, intentions and their recognition 123
represent, for the individual, a means of accessing the self, others and the environment; they are therefore studied 124
at all of these levels. 125
Despite the interest in these methods, the impossibility of directly assessing mental states has an impact on their 126
consideration, which affects the way in which they are studied and evaluated (Tuyttens et al., 2021) and the need 127
to improve the feasibility, reliability and validity of these methods is regularly outlined (Broom, 2011; Tuyttens et 128
al., 2021). This is true of human studies and therefore even more limiting for non-human animals. Volpato (2009) 129
highlighted that knowing that animals are sentient could have an influence on scientific observations (i.e. what is 130
observed and how). In other words, to be able to evaluate intentions in other species, one must first have a 131
representation of these species that allows it. To this first level of representation must be added that of the 132
“intention” itself. To be able to identify an intention, we need to know what is intentional and what is not. Thus, 133
depending on the discipline that studies it, intentions can be studied and defined very differently. This diversity is 134
all the more marked in philosophical reflections on the nature of intentions (see Bratman, 1987; Husserl, 1901; 135
Setiya, 2018 and many others). In short, intention is one of those research objects that reveal as much about those 136
who study them as those who are the subjects of study. 137
As the study of intentions is central to assessing the mental states but limited by the fact that it takes different 138
approaches due to the discrepancy of theoretical and experimental frameworks, we choose to investigate among 139
the larger possible set of disciplines whether intentions are truly considered in non-human animals. We address 140
the following three questions: 1. Is the concept of intention studied? 2. By whom (i.e. which disciplines/fields)? 3. 141
How is it studied, i.e. by which approaches, and through which questions? 142
To that end, we developed an innovative method articulated on three steps based on the hypothesis that studying 143
the scientific literature will provide insights allowing a better understanding of the theoretical frameworks 144

4
currently adopted by the academics (Mukherjee et al., 2022). In the first step, we investigated the evolution of the 145
study of intentions in the academic literature through a general bibliometric analysis. Based on our first results, 146
we gathered a relevant corpus on the study of intentions, but most of the articles focused on humans. The second 147
step was to create a representative corpus of studies on intentions in non-human animals, i.e. also including work 148
that deals with intentions without mentioning them. As it was not possible to focus directly on articles on non-149
human animals, we identified the terms used to study intentions in order to capture those related to non-human 150
animals. To this end, we worked on the co-occurrence network of authors’ keywords (Donthu et al., 2021; 151
Mukherjee et al., 2022), which reflect communities of questioning (i.e. scientific approaches). This method 152
allowed us to select relevant approaches and associated keywords. In the third step, which was to understand how 153
the intentions of non-human animals are studied, we recreated a new corpus based on the keywords identified in 154
these relevant approaches. Finally, we analysed it by a co-occurrence network of authors' keywords. Indeed, as 155
has been shown for other concepts (Aria et al., 2021; Donthu et al., 2021; Jaakkola, 2020; Mukherjee et al., 2022), 156
we expected through these co-occurrence networks to identify current issues and reveal gaps in the study of 157
intentions in non-humans, and to take a first step towards opening up the current theoretical framework in ethology 158
as well as for proposing original opportunities for future research. 159
Methods 160
As each step of our protocol depends on the results of the previous one, the following sections develop the 161
methods for each step chronologically and, where necessary, refer to the relevant part of the results. 162
1. Step 1: General bibliometric analysis 163
This first step of our method was to investigate the evolution of the study of intentions in the academic literature 164
and create a relevant corpus on the study of intention. 165
a. Choice of database 166
Non-human animals are studied in different disciplines, either directly (where the non-human is the subject of 167
study) or indirectly (where the non-human is a model for understanding human processes; that is, the study of non-168
human cognitive abilities made to better understand human cognition from an evolutionary and developmental 169
perspective). Thus, to create a corpus representative of the variety of research related to intentions in non-human 170
animals, two multidisciplinary databases were targeted and compared: Web of Science (WoS) and Scopus. 171
According to Chadegani Arezoo et al. (2013), Scopus covers a wide range of disciplinary fields (20% more than 172
WoS) and a large number of journals. In addition, there are nine times more articles present only in Scopus than 173
in WoS (Chadegani Arezoo et al., 2013). The Scopus metadata format is the best adapted to use for the corpus 174
analysis as author keywords and automatically indexed keywords (i.e. keywords proposed by the database itself) 175
are not distinguished in the WoS when downloading metadata, unlike those of Scopus (Tancoigne et al., 2014). 176
Furthermore, we used the CorTexT platform (IFRIS and INRA, https://www.cortext.net/) to create the co-177
occurrence networks of author keywords. Indeed, CorTexT was created to quantitatively and qualitatively explore 178
bibliographic data and offers tools to explore relationships between concepts, scientific communities and more 179
(CorText Platform, 2023; examples of studies using CorText: Brás et al., 2017; Chavalarias and Cointet, 2013; 180
Deng and Xia, 2020; Malanski et al., 2021; Mesmoudi et al., 2015; Raimbault et al., 2016; Weisz et al., 2017). A 181
further advantage of CorText lies in its optimal management of metadata in 'Scopus RIS' format. The corpus was 182
therefore created with the Scopus database. We chose to restrict our analysis from 1990 to 2020 – from the year 183
of the oldest articles identified on intentions in Scopus to the last completed year – in order to ensure stability in 184
articles referenced throughout the bibliometric analysis. 185
b. Creation of the first corpus: intentions in scientific literature 186
To limit noise due to the verbal form (to intend) and its conjugation, but while remaining as exhaustive as 187
possible, the query focused on the word “intent*” in titles and keywords. Only English papers were selected. This 188
initial analysis led us to exclude from the query1 some non-relevant expressions using “intention” (“Intention To 189
1 Final query: TITLE ( "intent*" ) OR KEY ( "intent*" ) AND ( EXCLUDE ( PUBYEAR,2022) OR EXCLUDE
( PUBYEAR,2021) ) AND ( EXCLUDE ( EXACTKEYWORD, "Intention To Treat Analysis" ) OR EXCLUDE (
EXACTKEYWORD, "Sensory Deprivation (Intentional)" ) OR EXCLUDE ( EXACTKEYWORD, "Intentional
Contamination" ) OR EXCLUDE ( EXACTKEYWORD, "Intentional Sampling" ) OR EXCLUDE (
EXACTKEYWORD, "Intended Dietary Use" ) OR EXCLUDE ( EXACTKEYWORD, "Intentional Feeding" ) OR
EXCLUDE ( EXACTKEYWORD, "Intentional Electromagnetic Interference" ) OR EXCLUDE (

5
Treat Analysis"; "Sensory Deprivation (Intentional)”; "Intentional Contamination"; "Intentional Samplin g"; 190
Intended Dietary Use and "Intentional Feeding"; “intentional electromagnetic interference”; “non-intentionally 191
added substances (nias)”). The obtained corpus is referred to hereafter as the “intent*” corpus. 192
2. Step 2: Identification of the scientific approaches of intentions (human and non-193
human animals) 194
The second step was to identify the different scientific approaches to studying intentions in order to reveal those 195
specific to the study of non-human animals as well as the terms associated with them. For this, we used the keyword 196
co-occurrence network as a means of identifying these distinct research communities, i.e. distinct scientific 197
questions (Aria et al., 2021; Mukherjee et al., 2022; Tancoigne et al., 2014). This method is known to allow the 198
identification of different dimensions of the concept of interest (Gauld and Micoulaud-Franchi, 2020) and the types 199
of questions developed to study it (Donthu et al., 2021; Jeanneaux et al., 2012). 200
a. Creation of the second corpus: intentions in “Agricultural & Biological Sciences” 201
Through our bibliometric work, we analysed publication trends over the last 30 years in order to determine the 202
period on which we would focus. We found that 2016 was a pivotal year in terms of the increase in the number of 203
articles on intentions. Because of this inflection, we decided to focus on articles published between 2016 and 2020, 204
on the assumption that they would reveal a greater diversity of approaches (see Result 1). Furthermore, limiting 205
the study to those four years ensures that the analysis is representative of current research issues. Then, due to the 206
lack of keywords enabling the search to be restricted to non-humans or to exclude humans, the diversity of 207
disciplines mentioned above and the lack of universal rules for referencing articles by keywords, an accurate focus 208
on non-human animals was not directly possible. Therefore, using the general “intent*” corpus, we decided to 209
develop a method to screen the literature in search of indices of the knowledge on non-human animals’ intentions. 210
After analysing the 20 Scopus subject areas involved in this corpus, we focused on the Scopus subject area 211
“Agriculture & Biological sciences”. This Scopus area covers a wide range of journals related to the study of non-212
human animals. The list provided by Scopus (last accessed February 2023) includes 31 151 journals. In addition, 213
as the same journal can be assigned to different Scopus areas, journals from other disciplines (such as psychology 214
and neuroscience, for example) can also be found under this label. In other words, focusing on the Scopus area 215
“Agricultural & Biological Sciences” did not exclude any scientific field that may work with non-human animals. 216
Thus, we argue that our corpus, focused on the Scopus area “Agriculture & Biological science”, is representative 217
of the diversity of study on the intentions of non-human animals. The final corpus, hereafter referred to as “Ag&B 218
intent* corpus”, consisted of 936 articles. 219
b. Selection of keywords analysed by co-occurrence network 220
In order to focus on the researchers’ specific research questions, which are better represented by the authors’ 221
keywords, we removed from our analysis the keywords automatically indexed by Scopus (Aria et al., 2021). The 222
200 most frequent author keywords were extracted (for more details see https://docs.cortext.net/lexical-extraction/ 223
on the CorText Platform). Then, while respecting idiosyncrasies, forms with spelling differences 224
(presence/absence of hyphen, plural/singular, British or American spelling differences such as “behaviour” and 225
“behavior”) were grouped. In order to build networks on notions related to intentions, all keywords related to the 226
study population and to the method (both those related to design and those related to data collection techniques) 227
were removed. This selection of the most frequent keywords had the effect of excluding from the corpus (and 228
therefore from the rest of the work) articles that did not contain any of these keywords. Thus, 715 articles (out of 229
936) were used to create the co-occurrence network of author keywords (Jeanneaux et al., 2012). 230
c. Keyword co-occurrence networks corpus “Ag&B intent*” and selection of specific 231
approaches in studies of non-human animal intentions 232
The CorTexT Platform (IFRIS and INRA, https://www.cortext.net/) was used to create networks of keywords. 233
They were produced as follows: for each keyword, the sum of the number of co-occurrences with all other 234
EXACTKEYWORD, "Intentional Electromagnetic Interference (IEMI)" ) OR EXCLUDE ( EXACTKEYWORD,
"IEMI" ) OR EXCLUDE ( EXACTKEYWORD, "Second Intention Healing" ) OR EXCLUDE ( EXACTKEYWORD,
"Second-intention Healing" ) OR EXCLUDE ( EXACTKEYWORD, "Second Intention Wound Healing" ) OR
EXCLUDE ( EXACTKEYWORD, "Second-intention Wound Healing" ) OR EXCLUDE ( EXACTKEYWORD,
"NIAS" ) OR EXCLUDE ( EXACTKEYWORD, "Non-intentionally Added Substances" ) OR EXCLUDE (
EXACTKEYWORD, "Non-intentionally Added Substances (NIAS)") OR EXCLUDE ( EXACTKEYWORD, "Non
Intentionally Added Substances (NIAS)" ))

6
keywords was calculated (node weight) and then each keyword was associated with the 5 keywords with which it 235
co-occurred the most, according to the proximity measure (edges). This method is a distributional measure which 236
counts the co-occurrence for one term with all other terms in the same context. Thus, the closer the nodes are, the 237
more they co-occurred in a related context. Communities of terms are proposed, based on the classical Louvain 238
resolution (Blondel et al., 2008). This algorithm optimises the modularity of each community (Blondel et al., 239
2008). Each keyword belongs to only one community. The prevalence of each keyword in a community is given 240
as the weight measure. Each community is named by the two nodes with the highest degree, which corresponds to 241
the centrality measure on the CorText Platform (Brás et al., 2017). 242
The communities named by the two nodes were more deeply analysed in the objective to be used as highlighters 243
of the scientific approaches in non-human animals that may be backboned by the concept of intention. Three of 244
these were selected because they were formed by keywords used in studies on non-human animals (see Results 245
2.). 246
3. Step 3: Identification of the scientific approaches of non-human animals’ 247
intentions 248
The previous step identified various specific scientific approaches to the intentions of non-human animals and 249
the terms used in them. In this stage, the aim was to obtain a representative view of the way in which intentions 250
are currently studied in non-human animals. To this end, the method developed here was to select a corpus solely 251
focusing on non-human animals and representative of the diversity of current research on the subject of intentions. 252
a. Creation of the third corpus: intentions in non-human animals 253
In order to select a corpus representative of the diversity of studies on non-human animal intentions, we decided 254
to build it from the terms identified in the previous step. However, in order to limit the noise of articles unrelated 255
to intentions, and to retain the information carried by co-occurrence, the queries were systematically built around 256
the association of two keywords. Thus, for each community, the highest-weighted keywords were selected. Then, 257
based on their combinations (Table 1), three queries2 (one per cluster, see Result 2) were created for articles in 258
English and for the period 2016-2020. The resulting corpus contained a total of 1022 articles (one article was 259
identified by two of the three queries, but counted only once in the final corpus). Of these articles, 111 were 260
identified as focusing on non-human animals (see supplementary data Table 8). These 111 articles form the corpus 261
on non-human animals, hereafter referred to as the “Non-human animals’ intentions” corpus. 262
b. Keyword co-occurrence networks corpus “Non-human animals’ intentions” 263
As in the previous step, the authors’ 200 most frequently used keywords (except for those related to the methods 264
and the study population) of the “non-human animals’ intentions” corpus were selected. Given the size of the 265
2 Query Community 1: ( AUTHKEY ( "shared intentionality" ) AND AUTHKEY ( "cooperation" ) ) AND (
: LIMIT-TO ( PUBYEAR , 2020 ) OR LIMIT-TO ( PUBYEAR , 2019 ) OR LIMIT-TO ( PUBYEAR , 2018 )
OR LIMIT-TO ( PUBYEAR , 2017 ) OR LIMIT-TO ( PUBYEAR , 2016 ) ) AND ( LIMIT-TO ( LANGUAGE ,
"English" ) ); Query Community 2: ( AUTHKEY ( "referential communication " ) AND AUTHKEY ( "Social
cognition " ) ) OR ( AUTHKEY ( " referential communication " ) AND AUTHKEY ( "domestication" ) ) OR (
AUTHKEY ( " Social cognition " ) AND AUTHKEY ( " domestication" ) ) AND ( LIMIT-TO ( PUBYEAR , 2020
) OR LIMIT-TO ( PUBYEAR , 2019 ) OR LIMIT-TO ( PUBYEAR , 2018 ) OR LIMIT-TO ( PUBYEAR , 2017
) OR LIMIT-TO ( PUBYEAR , 2016 ) ) AND ( LIMIT-TO ( LANGUAGE , "English" ) ); Query Community
3: (AUTHKEY ( "gesture" ) AND AUTHKEY ( "vocalization" )) OR (AUTHKEY ( " language evolution" ) AND
AUTHKEY ( "vocalization" )) OR (AUTHKEY ( " language evolution" ) AND AUTHKEY ( " intentionality" )) OR
(AUTHKEY ( " language evolution" ) AND AUTHKEY ( " flexibility" )) OR (AUTHKEY ( " language evolution" )
AND AUTHKEY ( " language" )) OR (AUTHKEY ( " language evolution" ) AND AUTHKEY ( " gesture" )) OR
(AUTHKEY ( " flexibility" ) AND AUTHKEY ( " gesture" )) OR (AUTHKEY ( " flexibility" ) AND AUTHKEY ( "
language" )) OR (AUTHKEY ( " flexibility" ) AND AUTHKEY ( " intentionality" )) OR (AUTHKEY ( " flexibility"
) AND AUTHKEY ( " vocalization" )) OR (AUTHKEY ( " vocalization " ) AND AUTHKEY ( " language " )) OR
(AUTHKEY ( " vocalization " ) AND AUTHKEY ( " intentionality " )) OR (AUTHKEY ( " intentionality " ) AND
AUTHKEY ( " gesture " )) OR (AUTHKEY ( " intentionality " ) AND AUTHKEY ( " language" )) OR (AUTHKEY
( " gesture " ) AND AUTHKEY ( " language" )) AND ( LIMIT-TO ( PUBYEAR,2020) OR LIMIT-TO (
PUBYEAR,2019) OR LIMIT-TO ( PUBYEAR,2018) OR LIMIT-TO ( PUBYEAR,2017) OR LIMIT-TO (
PUBYEAR,2016) ) AND ( LIMIT-TO ( LANGUAGE, "English" ) )

7
corpus and the low frequency of keywords, only the 100 most frequent ones were kept. Finally, based on this list, 266
a keyword co-occurrence network was obtained by following the same method as that described in section 3.2. 267
Table 1: Keyword combinations of the queries used to create the “Non-human animal’s intentions” corpus. 268
Results 269
1. General bibliometric analysis: intentions are studied in many fields 270
The following results were obtained on the corpus “intent*” corpus (query focusing on the word “intent*” in 271
the title and keywords and without the identified irrelevant expressions, see Method 2.b). The prevalence of studies 272
on the concept of intention, illustrated in Fig. 1, shows the number of publications on the concept for a given year, 273
weighted by the total number of publications in the Scopus database. This weighting compensates for the 274
exponential growth of scientific production (Bornmann and Mutz, 2014) to reflect the actual evolution of the 275
proportion of studies on the concept of intention. The number of articles studying the concept of intention has 276
increased sevenfold in the last 30 years. An acceleration of this increase in publication is to be noted from 2015 277
onwards (Fig. 1). 278
Keywords selected
Combinations
Community 1
shared intentionality OR cooperation
shared intentionality AND cooperation,
Community 2
referential communication OR social
cognition OR domestication
referential communication AND social cognition,
social cognition AND domestication,
referential communication AND domestication,
Community 3
language evolution OR flexibility OR
vocalization OR intentionality OR
language OR gesture
gesture AND vocalization,
language evolution AND vocalization,
language evolution AND intentionality,
language evolution AND flexibility,
language evolution AND language,
language evolution AND gesture,
flexibility AND gesture,
flexibility AND language,
flexibility AND intentionality,
flexibility AND vocalization,
vocalization AND language,
vocalization AND intentionality,
intentionality AND gesture,
intentionality AND language,
gesture AND language

8
279
Fig. 1 Prevalence over time of the study of the concept of intention on the Scopus database from 1990 to 2020. 280
The query aimed to select the articles with the word « intent » and its derivatives in the keywords and titles 281
(“intent*” corpus). The prevalence is the number of selected articles out of the total number of articles in the 282
database per year. The increase in publications on intent is to be noted from 2015. 283
284
We then wanted to explore the disciplinary dynamics underlying this evolution. To do this, we looked at the 285
“subject area” analysis proposed by Scopus, with each subject area approximately reflecting a specific disciplinary 286
field. 287
The percentage of each discipline in the total corpus over time, presented in Fig. 2, illustrates the disciplinary 288
dynamics underlying the evolution of the “intent*” corpus. Three dynamics can be identified: the proportion of 289
the discipline in the total corpus that is stable (e.g. social sciences); that which decreases (e.g. medicine); and that 290
which increases (e.g. computer science). The data in Fig. 2 alone does not indicate whether the total number of 291
articles in a given discipline follows the same dynamic. The two types of information presented in Table 2 – the 292
weight of each discipline in the total corpus (total proportion in %) and the evolution of the proportion of each 293
discipline over time (multiplier coefficient) – complete the graph. When combined, this information sheds light 294
on the general dynamics of the disciplines. The scientific output on the concept of intention increased in only ten 295
of the thirty disciplines considered since 1990. The global increase seems to be due to computer science (Multiplier 296
Coefficient: 8.8 and Total Proportion: 15%). Two other disciplines seem to have an impact on the global evolution 297
of the study of intentions; namely, decision science and engineering. Decision science is the discipline that has 298
increased the most (Multiplier Coefficient: 12.9) but its total percentage remains low (4%). Engineering represents 299
a significant proportion of the corpus (9%), with its percentage having doubled. In contrast, the proportions of the 300
major disciplines in our corpus, Social sciences and Medicine and Psychology, have decreased (Multiplier 301
Coefficient respectively: 0.9; 0.5 and 0.4) but they still represent a significant share of the total corpus 302
(respectively: 18%; 11% and 7%). Agricultural and Biological Sciences represent less than 2% of the total corpus 303
but their share has tripled in the total corpus (Table 2). 304
305

9
306
Fig. 2 Evolution of the prevalence of Scopus subject area in the “intent*” corpus, per year. Three dynamics can 307
be identified: the proportion of the discipline in the total corpus that is stable (e.g. social sciences); that which 308
decreases (e.g. medicine); and that which increases (e.g. computer science). 309
310
Table 2: Evolution of the proportion of each discipline between 1990 and 2020 in the "intent*" corpus. 311
Intent*
MC a
Total proportion b
DECISION SCIENCE
12.9
4%
COMPUTER SCIENCE
8.8
15%
ECONOMICS, ECONOMETRICS AND FINANCE
3.7
4%
AGRICULTURAL AND BIOLOGICAL SCIENCES
3.3
2%
ENVIRONMENTAL SCIENCE
3.0
5%

10
EARTH AND PLANETARY SCIENCES
2.3
1%
MATHEMATICS
2.2
3%
ENGINEERING
2.0
9%
PHYSICS AND ASTRONOMY
1.3
1%
CHEMICAL ENGINEERING
1.1
0%
SOCIAL SCIENCES
0.9
18%
MATERIALS SCIENCE
0.7
1%
BIOCHEMISTRY, GENETICS AND MOLECULAR BIOLOGY
0.7
2%
BUSINESS, MANAGEMENT AND ACCOUNTING
0.7
2%
HEALTH PROFESSIONS
0.6
1%
PHARMACOLOGY, TOXICOLOGY AND PHARMACEUTICS
0.6
1%
IMMUNOLOGY AND MICROBIOLOGY
0.5
0%
ARTS AND HUMANITIES
0.5
5%
MEDICINE
0.5
11%
PSYCHOLOGY
0.4
7%
NURSING
0.4
2%
CHEMISTRY
0.3
0%
DENTISTRY
0.2
0%
ENERGY
NA
3%
MULTIDISCIPLINARY
NA
1%
NEUROSCIENCE
NA
1%
UNDEFINED
NA
0%
VETERINARY
NA
0,3%
a MC: Multiplier Coefficient share of discipline in the annual corpus (SU)
b Total share: total share in the corpus (all papers between 1990-2020)
312
2. Identification of scientific approaches of intentions in the selected corpus 313
“Ag&B intent*” 314
From this point onwards, the work focused on the period 2016-2020, which corresponds to the period of 315
increased publication on intentions. As previously stated, we believe that this period is the most suitable to explore 316
a maximum of different scientific approaches while reflecting what is currently happening in research. 317
As previously explained (see Method 1.b), a direct focus on non-human animals was not possible. To 318
circumvent this limitation, we concentrated next on the corpus based on the Scopus area “Agricultural & Biological 319
Sciences”. This garnered 936 articles (“Ag&B intent*” corpus). All results presented in this section, except for the 320
“Frequency %” which was calculated manually, are from the CorText Platform. 321
a. Author keywords 322
Of the 200 most frequent keywords in the corpus, 149 were retained after deleting keywords relating to the 323
methods and the population studied. The most frequent keywords appeared in 84 articles, which represented 9% 324
of the total corpus (Table 3), suggesting a great variability in terms of notions involved in the study of intentions. 325
Most of the terms (139) were found in fewer than 10 different articles. 326
Table 3: Fifteen most frequent author keywords of the “Ag&B intent*” corpus after cleaning. Corpus: Scopus 327
subject area "Agricultural & Biological Sciences". Extraction via the CorText Platform (IFRIS and INRA, 328
https://www.cortext.net/). 329
Author keyword
Frequency a
Frequency % b
Purchase intention
84
9%
Theory of planned behaviour
69
7%
Intention
47
5%
Attitude
39
4%
Behavioural intention
30
3%
Consumer behaviour
28
3%
Consumer
20
2%
Organic food
18
2%

11
Purchase intent
17
2%
Intentionality
16
2%
Trust
14
1%
Emotion
13
1%
Satisfaction
12
1%
Food safety
10
1%
Subjective norm
10
1%
a Occurrences, equivalent to the number of articles
b Frequency relative to the total corpus
330
b. Co-occurrence networks of the notions involved in the study of intentions 331
In the network, each node represents the main form of the keyword. Its size is the sum of its co-occurrences 332
(see Method 3.b) and is given as its weight. The sum of the co-occurrences was calculated only for the 149 most 333
frequent keywords, which means that only the co-occurrences between the words in this list are considered for the 334
calculation of the weight of the nodes. As the frequency was calculated on the totality of the keywords in the entire 335
corpus, the ten high-weighted keywords of the co-occurrence network (Table 4) are not systematically the most 336
frequent ones in the corpus. These ten keywords are associated with communities that approach intentions from 337
the viewpoint of human behaviours and consumption (Fig. 3). The two most important keywords (high-weighted), 338
“intention” and “behaviour”, co-occur twice as often as the third heaviest. It is interesting to observe that they are 339
not at the centre of the network (Fig. 3). 340
Table 4: The ten high-weighted author keywords in the co-occurrence network “Ag&B intent*” corpus after 341
cleaning. Corpus: Scopus subject area "Agricultural and Biological Sciences". Extraction via the CorText 342
Platform (IFRIS and INRA, https://www.cortext.net/). 343
Author keyword
Weight a
Frequency b
Intention
522
47
Behaviour
512
9
Consumer behaviour
262
28
Perception
261
6
Purchase intention
224
84
Attitude
219
39
Theory of planned behaviour (TPB)
209
69
Consumer
194
20
Emotion
116
13
Education
82
5
a Co-occurrence sum
b Occurrences, equivalent to the number of articles
344

12
345
Fig. 3 Co-occurrences of the 200 most frequent author keywords on the study of intentions (2016 to 2020), “Ag&B intent*” corpus.
Corpus: Scopus area “Agricultural and Biological sciences” plotted through the CorText Platform (IFRIS and INRA,
https://www.cortext.net/). The thickness of the edge represents the number of co-occurrences: the thicker the edge, the more the
two related words co-occur. For a given node, its position in the network is calculated relative to the position of all other nodes.
Thus, the length of the edges can be interpreted as the proximity of two words, i.e. the shorter the edge, the more the words occur
in the same context. Measure: distributional. Threshold: top-5 neighbours. TPB: Theory of Planned Behaviour.

13
The geometric organisation of the network (Fig. 3) can be interpreted as follows. Based on the Louvain 346
resolution (with a resolution parameter of 1), 11 stable communities (modularity: 0.75) were detected (Table 5 and 347
Fig. 3). Each community is named by its two high-weighted keywords. They are represented by coloured circles 348
(Fig. 3). The size of each circle is proportional to the number of articles associated with the community (Table 6). 349
All the communities have a high density (Table 5), which means that each keyword co-occurred with almost all 350
the keywords of its community. This is particularly true for the “emotion & purchase intent”; “climate change & 351
adaptation” and “language & gesture” communities. Conversely, the “service quality & behavioural intention” 352
community seems less homogenous. 353
In the author keywords co-occurrence network (Fig. 3), the communities are organised into three meta-354
communities (i.e. a spatial grouping of several communities; for details, see Table 5). The main meta-community, 355
in terms of the number of communities belonging to it (referred to afterwards as meta-community 1), is located at 356
the bottom of the network. It is composed of seven clusters, all related to consumption and consumer behaviour. 357
It is interesting to note that the keyword “animal welfare” is linked to the keywords “Knowledge”, “Education” 358
and “Adoption” and not to the keywords of animal behaviour studies. The second meta-community (meta-359
community 2, the second one up the network) concerns risks linked with the production and consumption of food. 360
The co-occurrence network analysis revealed the most frequent derivatives (expression or word) associated with 361
the word "intent*". In total, 14 different derivatives were found (Table 5). 362
For a given keyword, the betweenness was normalized between [0;1]. The keywords with the highest 363
betweenness centrality are the most central keywords from the point of view of the geometrical organisation, i.e. 364
they are keywords at the intersection of the shortest paths between the other nodes. The ten keywords with the 365
highest normalised betweenness centrality are presented in Table 5; all these keywords belong to the meta-366
communities 1 or 2. Finally, the third meta-community (meta-community 3, at the top of the network) is the only 367
one with keywords related to studies on non-human animals’ behaviours. It is composed of three approaches: 368
language and gesture; shared intentionality and cooperation; and referential communication and domestication. 369
This meta-community is linked to the others only by the keyword “communication” through the community 370
“language & gesture”. Given the centrality of the keyword “communication” in the network and its low 371
betweenness centrality value, meta-community 3 represents relatively few paths between nodes. In other words, 372
there are few connections between the meta-community 3 and the other two (Table 5 and Fig. 3). 373

14
Table 5: Communities, meta-communities and betweenness centrality measures of the co-occurrence network of the 200 most frequent author keywords on the study of 374
intentions (2016 to 2020). Corpus: Scopus area “Agricultural and Biological Sciences” plotted by CorText Platform (IFRIS and INRA, https://www.cortext.net/). Community 375
measure: Louvain resolution. 376
Meta-communities a
Communities b
Density
Intent* Forms
Normalized Betweenness Centrality.10-3
Network c
Community d
1
intention & behaviour
0.78
Intention
Perception
7
Intention to use
ecotourism & brand image
0.86
Repurchase intention
Ecotourism
8
Consumption intention
Revisit intention
Entrepreneurial intention
sensory evaluation & meat
0.78
Sensory evaluation
37
Sensory evaluation
37
Meat
33
Food consumption
25
Sustainability
23
purchase intention & consciousness
0.78
Purchase intention
Purchase intention
15
Purchase intention
15
emotion & purchase intent
0.92
Purchase intent
Consumer acceptance
33
Consumer acceptance
33
Purchase intent
23
Emotion
16
education & sugar
0.85
Animal welfare
13
service quality & behavioural intention
0.71
Behavioural intention
Marketing
14
Marketing
14
Turnover intention
2
climate change & adaptation
0.97
Climate change
17
Climate change
17
3
language & gesture
0.92
Intentionality
Communication
2
Intentional
shared intentionality & cooperation
0.86
Shared intentionality
Cooperation
1
referential communication & domestication
0.88
Intentional communication
Domestication
0.4
a Spatial organisation
b Louvain Resolution
c 10 keywords with the highest betweenness centrality.
d Keywords with the highest betweenness centrality for each cluster.
377

15
Table 6: Community names and number of articles associated with each community from the co-occurrence 378
network author keywords “Ag&B intent*” corpus. The name of the communities consists of the two high-weighted 379
keywords. Corpus: Scopus subject area "Agricultural and Biological Sciences". Extraction via the CorText 380
Platform (IFRIS and INRA, https://www.cortext.net/). 381
Community name
Article count
purchase intention & consciousness
208
emotion & purchase intent
160
ecotourism & brand image
51
sensory evaluation & meat
42
language & gesture
37
climate change & adaptation
32
education & sugar
126
shared intentionality & cooperation
13
referential communication & domestication
15
job satisfaction & turnover intention
10
service quality & convenience
21
382
3. The focus on non-human animals’ intentions 383
As explained in the method section (see Method 4.b), 111 papers (see supplementary data, Table 9) on non-384
human animal intention were selected. The same method of analysis was used. All the results presented in this 385
section, except for the “Frequency %” which was calculated manually, are from the CorText Platform. 386
a. Author keywords 387
The most frequent keyword (“Language evolution”) appeared in 66 articles, which represented 59% of the total 388
corpus, suggesting a lower variability in terms of notions involved in the study of intention in non-humans than in 389
humans. The corpus on non-human animals built upon 6 keywords, all linked to communication except one, 390
“Intentionality” (Table 7). This theme is also central in the spatial organisation of the network: 7 of the 10 high-391
weighted keywords are related to it (Table 7). 392
Table 5: The ten most frequent (non-grey cells) and the ten high-weighted (non-grey cells) author keywords in the 393
co-occurrence network. “Non-human animals’ intentions” corpus after cleaning. Extraction via the CorText 394
Platform (IFRIS and INRA, https://www.cortext.net/). 395
Author keyword
Frequency a
Frequency % b
Weight C
Language evolution
66
59%
258
Gesture
19
17%
108
Communication
18
16%
150
Vocalisation
16
14%
189
Intentionality
14
13%
67
Animal communication
12
11%
107
Language
10
9%
223
Social cognition
10
9%
49
Referential communication
7
6%
34
Syntax
7
6%
33
Evolution
3
3%
152
Cognition
5
4%
101
Speech
4
4%
67

16
a Number of occurrences, equivalent to the number of articles
b Frequency relative to the total corpus
c Co-occurrence sum
396
b. Co-occurrence networks of the notions involved in the study of intentions in non-397
human animals 398
The Louvain resolutions are stable, with 10 communities (modularity: 0.77) organised in two meta-399
communities (Fig. 4 and Table 8). As for the previous network, the size of the colour circles is proportional to 400
their number of articles and the name of the community is given by the two high-weighted keywords. As in the 401
previous network, all communities have a high density, with the highest density in the “teaching & tradition” 402
community (Table 8). Conversely, the “service quality & behavioural intention” community seems less 403
homogenous. 404
The main meta-community (meta-community 1), composed of nine of the ten communities, is related to 405
humans either through the origin of human language or through the comparison with humans. The ten keywords 406
with the highest normalized betweenness centrality are presented in Table 8; they all belong to meta-community 407
1, which organises the network around it. In addition, eight of the ten central keywords belong to three 408
communities: “mirror neuron and language”; “flexibility & meaning” and “human-animal interaction & 409
domestication” (Table 8). These communities contain keywords related to theories of language origin 410
(“multimodal”, “combinatoriality”, “compositionality”), to the comparison between humans and apes 411
(“comparative psychology”, “referential communication”), and to evolutionary theory (“language evolution”, 412
“language development”). The community “human-animal interaction & domestication” is also related to the 413
neurophysiology that supports language (“broca area”, “prefrontal cortex”) (Fig. 4). Meta-community 1 is 414
organised around two axes: one from eusociality to sociability and the other from audition to language. The first 415
axis is organised (in this order) from “self-domestication & diseases”, “human-animal interaction & 416
domestication”, “mirror neuron & language” to “behavioural flexibility & social context”. The second axis goes 417
from “comparative cognition & auditory” to “handedness & cultural evolution” via “brain evolution & cultural 418
evolution”, “human-animal interaction & domestication” and “flexibility & meaning” (Fig. 4). These two axes are 419
organised around the keyword “gesture”, the most central one in this network (Normalised betweenness centrality: 420
147.10-3). The “duets & antiphony” community is the only one that does not really lie on these two axes. The 421
second meta-community is that of “teaching & tradition”, and is composed of a single community, linked to the 422
other only by “intentionality” (normalized betweenness centrality: 101.10-3, Table 8). All the keywords of this 423
community are associated with transmission between individuals. It is interesting to note that the two communities 424
in which the notion of emotions appears are “self-domestication & disease” and “teaching & tradition”. 425

17
Table 6: Communities, meta-communities and betweenness centrality measures of the co-occurrences network of the 100 most frequent author keywords on the study of 426
intention in non-human animals (2016 to 2020). Corpus: “Non-human animals’ intentions” corpus plotted by CorText Platform (IFRIS and INRA, https://www.cortext.net/). 427
Community measure: Louvain resolution. 428
429
Meta-communities a
Communities b
Density
Normalized Betweenness Centrality.10-3
Network c
Community d
1
behavioural flexibility & social context
0.88
Behaviour
136
Behaviour
136
human-animal interaction & domestication
0.82
Social cognition
137
Social cognition
137
Referential communication
120
mirror neuron & language
0.83
Speech
91
Speech
91
Evolution
90
Communication
86
flexibility & meaning
0.84
Gesture
147
Gesture
147
Meaning
142
Intentionality
101
self-domestication & disease
0.84
Self-domestication
66
comparative cognition & auditory
0.94
Animal cognition
106
Animal cognition
106
handedness & manipulation
0.95
Language origin
11
brain evolution & cultural evolution
0.86
Primate communication
83
antiphony & duets
0.89
Human language
19
2
teaching & tradition
0.98
Imitation
11
a Spatial organisation
b Louvain Resolution
c 10 keywords with the highest betweenness centrality.
d Keywords with the highest betweenness centrality for each cluster.
430

18
431
Fig. 4 Co-occurrences of the 100 most frequent author keywords on the study of intention (2016 to 2020). Corpus: "Non-human
animals’ intentions” corpus plotted through the CorText Platform (IFRIS and INRA, https://www.cortext.net/). The thickness of the
edge represents the number of co-occurrences: the thicker the edge, the more the two related words co-occur. For a given node, its
position in the network is calculated relative to the position of all other nodes. Thus, the length of the edges can be interpreted as the
proximity of two words, i.e. the shorter the edge, the more the words occur in the same context. Measure: distributional. Threshold:
top-5 neighbours.

19
Discussion 432
In this paper, we hypothesised that the identification in the academic literature of current scientific approaches 433
to intention, and the gaps between them, might allow for a discussion of the current boundaries of the theoretical 434
and experimental framework of ethology. This would then open new avenues for exploring the intentions of non-435
human animals. Thus, by developing a “step-by-step” bibliographical method, we identified 111 articles on the 436
intentions of non-human animals that are representative of the current studies. Their analysis revealed 10 different 437
scientific approaches to the concept of intention. In the following section, we discuss the results obtained and their 438
limitations, following each step of our method. 439
Our work is based on the multidisciplinary Scopus database, which – despite a significant representation of 440
social sciences and humanities – remains limited. For example, with the same query we found approximate 12 000 441
articles on Scopus compared to 20 000 articles on PsycINFO. However, in order to understand the theoretical 442
environment of ethology researchers (i.e. the knowledge on which they base their own work), it makes sense to 443
focus on the databases used by these communities, rather than trying to achieve exhaustiveness. Our bibliometric 444
analysis revealed an increase in interest in intentions within the global academic literature. This interest increased 445
particularly from 2005, with an acceleration in 2015. Further examination of the underlying disciplinary dynamics 446
revealed that the increase in the number of studies can be caused by two different dynamics: an increase of interest 447
in the concept by a discipline already working on it; and the emergence of new disciplines. The increase in 448
publications on intentions seems to be driven by only six disciplines: computer science, decision science, 449
engineering, social science, medicine and psychology. Two dynamics are identifiable here: the emergence of new 450
disciplines (computer science, decision science and engineering); and increased interest from older ones (social 451
science and medicine). In other words, the concept of intention is not more studied in general but new disciplines 452
have taken an interest in it, while in older disciplines, interest has barely increased or has even decreased. These 453
results could support the idea that intentions per se is a useful concept for developing new disciplines and/or new 454
scientific questions. Thus, the exploration of intentions is a way to bring a new epistemological lens to a field, as 455
outlined by Cartmill and Hobaiter (2019), who used intentions as a marker of a particular state of gesture as a 456
window into the minds of great apes. 457
To assess how intentions are studied in non-human animals, we focused on the Scopus area "Agricultural & 458
Biological Sciences" which covers a wide range of journals related to the study of non-human animals. In the list 459
provided by Scopus (last accessed February 2023), this Scopus area includes 31,151 journals. As the same journal 460
can be assigned to different Scopus areas, journals from psychology and neuroscience, among others, can be found 461
under this label. For example, Animal cognition is tagged under Agricultural & Biological Sciences and in 462
Psychology areas. Thus, we have drastically reduced the risk of exclusion of an entire research field. This allows 463
us to analyse our results as a picture of the study of the intentions of non-human animals in sciences. Secondly, 464
and again with the aim of maintaining a representative aspect of current studies, we chose to base our query on 465
author keywords and indexed keywords. This method allows us to identify those papers in which the authors have 466
used the terms intention or intentionality as well as those that deal with intentions without making it explicit. These 467
include, for example, articles that study behaviours that involve intentions or can be inferred to intentions and 468
those that focus directly on intentions but do not label them as such. We have termed the adaptation by authors of 469
terms used by ethologists to those commonly accepted in zoology, the "self-censorship hypothesis". This 470
phenomenon is already known in another field of ethology; researchers of the concept of emotion in past studies 471
of non-humans (de Waal, 2011) have notably used the term "emotional reactivity" instead of "emotion" (Boissy, 472
2021). In other words, we found that, based on keywords alone, it was impossible to identify articles on the study 473
of intentions in non-human animals for the following reasons: 1) this term is not always used for the study of 474
intentions and the behaviours from which these mental states are inferred; and 2) researchers do not always specify 475
the subject of the study (whether involving humans or not). To overcome these limitations, we have developed a 476
bibliometric strategy to narrow the field of investigation and to explore more deeply by focusing on areas in which 477
articles of non-human animals’ intentions per se can be found. 478
In the first obtained corpus, built on the keyword “intent*” and centred on the Scopus area “Agricultural & 479
Biological Sciences”, the study of intentions involves a great diversity of vocabulary: each keyword is used in few 480
articles. This suggests a fragmentation of research lines around numerous topics of interest carried by small 481
communities. Among these author keywords, it is interesting to note that the 15 most frequent author keywords 482
are not related to the behaviours to which an intention is generally inferred, i.e. those that serve to mark the 483
expression of an intention. On the contrary, these keywords are rather linked to the prediction of a behaviour by 484

20
studying the intentions that motivate it. These keywords can be divided into four categories: keywords related to 485
the subject who expresses the intention (“consumer”), the object of the intention (“organic food”, “food safety”), 486
the consequence of the intention (the subject of study) (“purchase intent / intention”, “attitude”, “behavioral 487
intention” and “consumer intention”) and finally what moderates and/or predicts the intention (“trust”, “emotion”, 488
“satisfaction” and “subjective norms”). In other words, we do not study intentions through their expression in 489
behaviours but instead we study the prediction of behaviours by speculating on intentions. 490
These results are confirmed by the co-occurrence network analysis. As proposed by several researchers (see 491
Mukherjee et al., 2022), co-occurrence networks can be used to identify research themes in a particular field, 492
scientific approaches to a theme (Aria et al., 2021; Tancoigne et al., 2014), and even for concept analyses (e.g. 493
Brás et al., 2017). Moreover, by focusing our analysis on the keywords used by the authors, we were able to 494
identify in detail the issues explored by the researchers in their work as well as the underlying theoretical and 495
methodological approaches. 496
The co-occurrence network of author keywords performed on the keywords “intent*” from the domain-based 497
corpus of Scopus area “Agricultural & Biological Sciences” revealed 11 distinct scientific approaches 498
(communities). Each of these communities has a high density, i.e. all the keywords co-occur with all the others. 499
Thus, each community reflects a coherent and homogenous research theme. The less dense communities 500
(“intention & behavior”, “sensory evaluation & meat”, “purchase intention & consciousness” and “service quality 501
& behavioural intention”) can be explained by the existence of research sub-themes. It would be interesting to 502
explore this dynamic in greater depth in future research. The organisation of this first network reveals that the main 503
themes on intentions in Scopus area “Agricultural & Biological Sciences” focus on consumer behaviours. In this 504
corpus, the main meta-community (the first from the bottom of the network, see Figure 3) is composed of 618 505
articles dealing with animal consumption or the impact (environmental, social, etc.) of the production of animal 506
products. This raises the question of where intention studies really stand in terms of the general direction of 507
scientific enquiry. 508
Let us focus on the specific topic of this first meta-community, such as welfare, which is particularly 509
meaningful. The author keyword “animal welfare” lies in the “education & sugar” community and co-occurs with 510
“education”, “meat”, “knowledge”, “adoption” (which refers to the adoption of behaviours), “food consumption” 511
and “belief”. None of these words are related to the study of non-human animals per se. Now, when we screen 512
another meta-community (the second one up the network, see Figure 3), that of “climate change & adaptation”, 513
we find 32 articles, all dealing with the notion of risk related to agricultural production. There are four types of 514
risk: socio-economic risk for farmers (“farmer decision making”), risk in the perception of agriculture (“risk 515
communication”, “risk perception”), environmental risks (“climate change”), and risks related to food 516
consumption (“obesity”). In brief, the studies of intentions in the set of articles that might best approach non-517
human animal intentions, i.e. Scopus area “Agricultural & Biological Sciences”, outline two meta-communities 518
closely linked to human behaviours or activities, with different approaches (cognitive, behavioural, social, 519
educative, etc.), and predominantly in relation to human production and consumption. These results, echoed by 520
the 15 most frequent keywords of the corpus, support that in this corpus, researchers are not studying intentions 521
per se by inferring them to behaviours, but are rather focusing on how intentions might be a good predictor of 522
specific behaviours (in this case, in particular consumption -related behaviours). In other words, the question seems 523
to be more related to the way in which intentions influence behaviour rather than whether they exist and how they 524
are expressed in the individual being studied. 525
Finally, the only meta-community which contains keywords fitting to non-human animal studies (the third at 526
the top of the network, see Figure 3), is the one composed around “language & gesture”, “shared intentionality & 527
cooperation” and “referential communication & domestication”. This meta-community contains only 65 papers of 528
our corpus (all species considered), representing only 9% of the articles indexed to the clusters (715 articles, of 529
the 937 total). Thus, despite the growing interest in recent years in the study of the concept of intention, only a 530
very small fraction is concerned with non-human animal intentions, if those 9% of the most accurate articles are 531
really linked to non-human animal intentions. 532
This meta-community is linked to the others only by the keyword “communication” (see Figure 3). If we accept 533
that those articles are really dealing with non-human animal intentions, it indicates that this latter is mainly 534
approached through communication pathways. Yet, we have seen that there are many other approaches, e.g. the 535
expression of behaviours oriented to a goal or a subject that follows the conditions of permanency and adaptability 536
(Burkart and van Schaik, 2020; Leavens et al., 2005). Moreover, all the keywords of the three communities are 537

21
linked to interaction through communication (e.g. “gestural communication”, “vocalisation”, “communication”, 538
“intentional communication”, “referential communication”), or cooperation (e.g. "cooperative breeding", 539
"cooperation"). Not only does communication seem to be the gateway to study intentions in non-human animals, 540
but it seems that the communication pathways are also indicators of the nature of the intention being studied, i.e. 541
intentions for or in a social interaction. This study of intentions in or through interactions implies a social context 542
and therefore a certain type of protocols. Thus, non-human animal intentions seem to be mostly reduced to a single 543
type of approach based on communication and social abilities, which can only involve a limited number of types 544
of mental and cognitive processes. 545
To explore further how non-human animals’ intentions are studied and following the concept of the “self-546
censorship hypothesis”, we created a second corpus based on keywords from the three communities of the third 547
meta-community, where the term related to the study of non-human animals is found. In this corpus, we found 548
only 111 articles related to non-human animals out of the 1022 articles of the corpus (11%). Based on the group 549
of 111 articles, the analysis of author keywords co-occurrence networks revealed 10 divergent scientific 550
approaches of intentions, 9 of which were in the same meta-community. This meta-community (Figure 2) is 551
organised around two axes: one going from eusociality to sociability and the other from hearing to language. 552
Again, these nine approaches are linked together by the term “gesture”. This ties in with the proposal of Cartmill 553
and Hobaiter (2019) to use gesture (and especially intentional gesture) to access the animal’s mind. The first axis 554
runs from "eusociality" to sociability ("social context", "mating behavior"). The intentions of non-humans seem 555
to be studied in relation to humans, through “human-animal interaction”, including “domestication”, but also an 556
evolutionary approach, whereby the animal is used to explore the origin of human cognitive capacities 557
("evolution", "language evolution", "comparative psychology", "speech evolution"). This axis also explores social 558
behaviours, in particular those of cooperation, up to and including reproductive behaviours. The second axis moves 559
through the subjects of study related to language, starting with the physiological capacity (hearing), then passing 560
through the cerebral structures (and their evolution, always in relation to humans), through gesture (and intention) 561
to arrive at language, with spoken language being the furthest point. These two axes allow us to understand how 562
intention is "dissected", and all the skills (physiological, cognitive and social) necessary for intentions to be 563
involved. It is interesting to note that the only community of this meta-community that is not on these two axes, is 564
the only one that is explicitly linked to the study of paradigms (“Antiphony & Duets” community). The latter 565
community, which does not belong to the previous meta-community, is related to “teaching and tradition” through 566
complex abilities and behaviours such as “tool use” and “social learning”. It differs from the other communities 567
not only by the subjects of study, but also by the temporality which is not only horizontal, but also vertical in that 568
it studies the persistence of behaviours over time (“culture”, “tradition”). Here, contrary to the results of previous 569
corpora (Scopus area “Agricultural & Biological Sciences”), intentions are indeed studied for themselves and no 570
longer as a behavioural predictor. Their study is fragmented around several major themes/questions: their origin, 571
with the specific question of the common ancestor with humans, their biological support, their expression through 572
a social context, and their transmission over time. Finally, in this corpus, the question of the existence of intentions 573
in the subject studied seems central. 574
These results need to be tempered by the limitations and biases that might be involved in creating the corpora. 575
As explained in the method section, it was not possible to focus directly on non-human animals and even less on 576
non-human animal intentions. Thus, during the various steps taken to obtain a corpus focused on non-human 577
animal intentions, choices had to be made (such as the database used, the keywords used, the writing of the queries, 578
etc.). It was therefore not possible to obtain an exhaustive corpus, and some areas of the study of intentions might 579
be missing from our corpora. Moreover, our work on Scopus area “Agricultural & Biological Sciences” only 580
focused on four years (2016 to 2020), in order to obtain a snapshot of the current scientific dynamics on these 581
issues. It might be interesting to compare our results with those of a similar study for other periods. However, 582
despite these limitations, in a first broad bibliometric analysis in which we explored the period from 1990 to 2020 583
and the keywords “intent”, “intend” and their derivates (data unpublished), we found a lower proportion of articles 584
on non-human animals than in the corpus focused on Scopus area “Agricultural & Biological Sciences”. Since the 585
biases on the selection of articles were not the same for these two stages, this confirms that the literature on the 586
intentions of non-human animals is still limited. Thus, despite these biases inherent in the bibliometric method, 587
the ten scientific divergent approaches identified surround, in some way, the concept of intention and its study in 588
non-human animals. 589
Furthermore, as proposed by Mukherjee et al. (2022), the use of co-occurrence network analyses of terms can 590
not only provide information about the organisation of current academic knowledge, but can also reveal gaps. 591

22
From this perspective, our results show that current scientific approaches to animal intentions are limited in terms 592
of the subjects of study (focused on the social context through the study of communication), but also from a 593
theoretical point of view, as this work highlights the predominance of approaches on humans in the studies of 594
intentions. On the one hand, our study reveals that humans, because of the origin of the concept of intention, are 595
used as the reference for what intentions are and how they are expressed. On the other hand, non-human animal 596
intentions are mainly studied as a means to better understand the origin of human intentions (and other cognitive 597
capacities). This indicates a lock-in that shadows the possibility of considering non-human animal intentions per 598
se. Our study provides tools to open the current theoretical and conceptual framework to intentions on non-human 599
animals. Indeed, the 10 divergent scientific approaches that we have revealed can be reasonably considered to 600
open the current ethological framework. By considering them, it would extend our ability to consider and study 601
the intentions of other species. We have already begun to test the opening of experimental approaches that they 602
allow. In so doing, we believe it will be possible to explore more broadly non-human mental states, which are still 603
difficult to access and assess. 604
605
Finally, in this article, we propose a first step towards a new theoretical framework for studying animal 606
intentions per se. Firstly, we provide the theoretical background and tools identified from the current academic 607
studies to develop new ways of considering animal intentions beside analysing the communication pathways. 608
Secondly, it could also be extremely enriching to put forward the hypothesis of "non-human intentions”. The point 609
here is that although a first definition and understanding of what constitutes intention must apparently be based on 610
human experience of these concepts, this does not impose a purely comparative approach. It might be possible to 611
start with a narrow human definition and then open it up to other forms of intentions, which could then be expressed 612
in other ways and be carried by other neuro-physiological processes, as is already the case for other cognitive 613
abilities (Mendl et al., 2011). The development of protocols would therefore focus on how to access and measure 614
an intention that cannot be directly conceived. Considering this assumption could have an impact on the design of 615
studies of animal intentions. 616
In conclusion, this study highlights the limitations of the current theoretical framework used to study non-617
human animals’ cognition, which does not allow for the full spectrum of non-human cognitive specificities. Our 618
research opens up a reflection on the theoretical framework for the study of intention in non-human animals, mental 619
skills that have so far received little attention in the scientific literature on non-human animals. The ten approaches 620
we identified throughout the bibliometric analysis can provide a conceptual framework for the hypothesis of “non-621
human intentions”. They could allow to avoid pre-conceptualising the nature of the intention and the form of its 622
expression and thus open the floor to innovative experimental approaches of intentionality in non-human animals. 623
The next step is to provide means to operationalise our conceptual framework. One way to do it is to test its 624
translation in different scientific communities. Actually, we created and conducted interdisciplinary workshops 625
based on our conceptual framework, which involved researchers in ethology, psychology, design science and 626
management who were able to collectively develop innovative protocols for studying intention in non-human 627
animals. The results of these workshops are reported and discussed in a following paper (Dauphiné-Morer et al., 628
Submitted). 629
Declarations 630
Ethical Approval 631
This declaration is not applicable. 632
Competing interests 633
The authors declare they have no conflict of interest to report. 634
Author contributions 635
Conceived and designed the study: all authors. Sourced the data: ALDM. Wrote the draft manuscript: ALDM. 636
Contributed to improving the text: all authors. 637
Funding 638
This work is part of a PhD thesis funded by the National Research Institute for Agriculture, Food and the 639
Environment (INRAE). 640
Availability of data and materials 641
The datasets are available from the corresponding author on request. 642
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which 643
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give 644
appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and 645
indicate if changes were made. The images or other third party material in this article are included in the article’s 646
Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included 647

23
in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds 648
the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this 649
licence, visit http://creat iveco mmons .org/licen ses/by/4.0/. 650
651

24
References 652
Ajzen, I., 2011. The theory of planned behaviour: Reactions and reflections. Psychology and Health 26, 1113–653
1127. https://doi.org/10.1080/08870446.2011.613995 654
Ajzen, I., 1991. The theory of planned behavior. Organizational Behavior and Human Decision Processes, 655
Theories of Cognitive Self-Regulation 50, 179–211. https://doi.org/10.1016/0749-5978(91)90020-T 656
Aria, M., Alterisio, A., Scandurra, A., Pinelli, C., D’Aniello, B., 2021. The scholar’s best friend: research trends 657
in dog cognitive and behavioral studies. Animal Cognition 24, 541–553. 658
https://doi.org/10.1007/s10071-020-01448-2 659
Baldwin, D.A., Baird, J.A., 2001. Discerning intentions in dynamic human action. Trends in Cognitive Sciences 660
5, 171–178. https://doi.org/10.1016/S1364-6613(00)01615-6 661
Blondel, V.D., Guillaume, J.-L., Lambiotte, R., Lefebvre, E., 2008. Fast unfolding of communities in large 662
networks. Journal of Statistical Mechanics: Theory and Experiment 2008, P10008. 663
https://doi.org/10.1088/1742-5468/2008/10/P10008 664
Boissy, A., 2021. Quand l’éthologie recourt à la psychologie humaine pour comprendre la sensibilité des 665
animaux, in: Baratay, É. (Ed.), Croiser Les Sciences Pour Lire Les Animaux, Homme et Société. 666
Éditions de la Sorbonne, Paris, pp. 159–166. 667
Boissy, A., Arnould, C., Chaillou, E., Désiré, L., Duvaux-Ponter, C., Greiveldinger, L., Leterrier, C., Richard, S., 668
Roussel, S., Saint-Dizier, H., Meunier-Salaün, M., Valance, D., Veissier, I., 2007. Emotions and 669
cognition: a new approach to animal welfare. Animal Welfare 7. 670
Bornmann, L., Mutz, R., 2014. Growth rates of modern science: A bibliometric analysis. Journal of the 671
Association for Information Science and Technology 28. 672
Brás, O.R., Cointet, J.-P., Cambrosio, A., David, L., Nunes, J.A., Cardoso, F., Jerónimo, C., 2017. Oncology 673
research in late twentieth century and turn of the century Portugal: a scientometric approach to its 674
institutional and semantic dimensions. Scientometrics 113, 867–888. https://doi.org/10.1007/s11192-675
017-2491-y 676
Bratman, M., 1987. Intention, Plans, and Practical Reason. Cambridge, MA: Harvard University Press. 677
Brentano, F., 1924. Psychologie vom empirischen Standpunkt, Meinier. ed. Liebnitz. 678
Broom, D.M., 2011. A History of Animal Welfare Science. Acta Biotheoretica 59, 121–137. 679
https://doi.org/10.1007/s10441-011-9123-3 680
Burkart, J.M., van Schaik, C.P., 2020. Marmoset prosociality is intentional. Animal Cognition 23, 581–594. 681
https://doi.org/10.1007/s10071-020-01363-6 682
Byrne, R.W., Cartmill, E., Genty, E., Graham, K.E., Hobaiter, C., Tanner, J., 2017. Great ape gestures: 683
intentional communication with a rich set of innate signals. Animal Cognition 20, 755–769. 684
https://doi.org/10.1007/s10071-017-1096-4 685
Cartmill, E.A., Hobaiter, C., 2019. Gesturing towards the future: cognition, big data, and the future of 686
comparative gesture research. Animal Cognition 22, 597–604. https://doi.org/10.1007/s10071-019-687
01278-x 688
Chadegani Arezoo, A., Salehi, H., Md Yunus, M., Farhadi, H., Fooladi, M., Farhadi, M., Ale Ebrahim, N., 2013. 689
A Comparison between Two Main Academic Literature Collections: Web of Science and Scopus 690
Databases. Asian Social Science 9, 18–26. https://doi.org/10.5539/ass.v9n5p18 691
Chavalarias, D., Cointet, J.-P., 2013. Phylomemetic Patterns in Science Evolution—The Rise and Fall of 692
Scientific Fields. PLoS ONE 8, e54847. https://doi.org/10.1371/journal.pone.0054847 693
CorText Plateform, 2023. Cortext digital plateform, Context, stakes and objectives. Cortext Digit. Plateform. 694
URL https://www.cortext.net/about-us/ (accessed 3.10.23). 695
CorText Plateform, 2022. Terms Extraction. Cortext Manager Documentation. URL 696
https://docs.cortext.net/lexical-extraction/ (accessed 5.12.22). 697
De Waal, F.B.M., 2011. What is an animal emotion? Annals of the New York Academy of Sciences 1224, 191–698
206. https://doi.org/10.1111/j.1749-6632.2010.05912.x 699
Deng, S., Xia, S., 2020. Mapping the interdisciplinarity in information behavior research: a quantitative study 700
using diversity measure and co-occurrence analysis. Scientometrics 124, 489–513. 701
https://doi.org/10.1007/s11192-020-03465-x 702
Dennett, D., 2009. Intentional Systems Theory, in: The Oxford Handbook of Philosophy of Mind. Oxford 703
University Press. https://doi.org/10.1093/oxfordhb/9780199262618.003.0020 704
Dennett, D.C., 1983. Intentional systems in cognitive ethology: The “Panglossian paradigm” defended. 705
Behavioral and Brain Sciences 6, 343–355. https://doi.org/10.1017/S0140525X00016393 706
Dickinson, A., Balleine, B.W., 2000. Causal cognition and goal-directed action, in: Heyes, C., Huber, L. (Eds.), 707
The Evolution of Cognition., Vienna Series in Theoretical Biology. MIT Press, Cambridge, MA, pp. 708
185–204. 709

25
Diener, E., Oishi, S., Lucas, R.E., 2009. Subjective Well-Being: The Science of Happiness and Life Satisfaction, 710
in: Lopez, S.J., Snyder, C.R. (Eds.), The Oxford Handbook of Positive Psychology. Oxford University 711
Press, pp. 186–194. https://doi.org/10.1093/oxfordhb/9780195187243.013.0017 712
Donthu, N., Kumar, S., Mukherjee, D., Pandey, N., Lim, W.M., 2021. How to conduct a bibliometric analysis: 713
An overview and guidelines. Journal of Business Research 133, 285–296. 714
https://doi.org/10.1016/j.jbusres.2021.04.070 715
Dretske, F., 2003. The intentionality of perception, in: John Searle. Cambridge University Press, pp. 1–15. 716
https://doi.org/10.1017/CBO9780511613999.007 717
Dretske, F., 1988. Explaining behavior: Reasons in a world of causes, Explaining behavior: Reasons in a world 718
of causes. The MIT Press, Cambridge, MA, US. 719
Dretske, F.I., 1980. The Intentionality of Cognitive States. Midwest Studies in Philosophy 5, 281–294. 720
https://doi.org/10.1111/j.1475-4975.1980.tb00408.x 721
Fischer, J., Zinner, D., 2011. Communication and Cognition in Primate Group Movement. International Journal 722
of Primatology 32, 1279–1295. https://doi.org/10.1007/s10764-011-9542-7 723
Gauld, C., Micoulaud-Franchi, J.-A., 2020. Analyse en réseau par fouille de données textuelles systématique du 724
concept de psychiatrie personnalisée et de précision. L’Encéphale 47, 341–347. 725
https://doi.org/10.1016/j.encep.2020.08.008 726
Genty, E., Heesen, R., Guéry, J.-P., Rossano, F., Zuberbühler, K., Bangerter, A., 2020. How apes get into and 727
out of joint actions: Shared intentionality as an interactional achievement. Interaction Studies 21, 353–728
386. https://doi.org/10.1075/is.18048.gen 729
Greiveldinger, L., Veissier, I., Boissy, A., 2011. The ability of lambs to form expectations and the emotional 730
consequences of a discrepancy from their expectations. Psychoneuroendocrinology 36, 806–815. 731
https://doi.org/10.1016/j.psyneuen.2010.11.002 732
Griffin, D.R., Speck, G.B., 2004. New evidence of animal consciousness. Animal Cognition 7, 5–18. 733
https://doi.org/10.1007/s10071-003-0203-x 734
Gupta, S., Sinha, A., 2019. Gestural communication of wild bonnet macaques in the Bandipur National Park, 735
Southern India. Behavioural Processes 168, 103956. https://doi.org/10.1016/j.beproc.2019.103956 736
Helfer, S.G., Elhai, J.D., Geers, A.L., 2015. Affect and Exercise: Positive Affective Expectations Can Increase 737
Post-Exercise Mood and Exercise Intentions. Annals of Behavioral Medicine 49, 269–279. 738
https://doi.org/10.1007/s12160-014-9656-1 739
Heyes, C., Dickinson, A., 1990. The Intentionality of Animal Action. Mind & Language 5, 87–103. 740
https://doi.org/10.1111/j.1468-0017.1990.tb00154.x 741
Husserl, 1901. Recherches logiques, Presses Universitaire de France 1959. ed. 742
Jaakkola, E., 2020. Designing conceptual articles: four approaches. AMS Review 10, 18–26. 743
https://doi.org/10.1007/s13162-020-00161-0 744
Jeanneaux, P., Aznar, O., Mareschal, S. de, 2012. Une analyse bibliométrique pour éclairer la mise à l’agenda 745
scientifique des « services environnementaux ». VertigO - la revue électronique en sciences de 746
l’environnement 12, 1–14. https://doi.org/10.4000/vertigo.12908 747
Krause, M.A., Udell, M.A.R., Leavens, D.A., Skopos, L., 2018. Animal pointing: Changing trends and findings 748
from 30 years of research. Journal of Comparative Psychology 132, 326–345. 749
https://doi.org/10.1037/com0000125 750
Kytö, E., Virtanen, M., Mustonen, S., 2019. From intention to action: Predicting purchase behavior with 751
consumers’ product expectations and perceptions, and their individual properties. Food Quality and 752
Preference 75, 1–9. https://doi.org/10.1016/j.foodqual.2019.02.002 753
Le Neindre, P., Dunier, M., Larrère, R., Prunet, P., 2018. La conscience des animaux, Quae. ed, Matière à 754
débattre et décider. Quae. 755
Leavens, D.A., Hopkins, W.D., 1998. Intentional communication by chimpanzees: a cross-sectional study of the 756
use of referential gestures. Developmental psychology 34, 813–822. https://doi.org/10.1037/0012-757
1649.34.5.813 758
Leavens, D.A., Russell, J.L., Hopkins, W.D., 2005. Intentionality as measured in the persistence and elaboration 759
of communication by chimpanzees (Pan troglodytes). Child Development 76, 291–306. 760
https://doi.org/10.1111/j.1467-8624.2005.00845.x 761
Malanski, P.D., Dedieu, B., Schiavi, S., 2021. Mapping the research domains on work in agriculture. A 762
bibliometric review from Scopus database. Journal of Rural Studies 81, 305–314. 763
https://doi.org/10.1016/j.jrurstud.2020.10.050 764
Mendl, M., Paul, E.S., Chittka, L., 2011. Animal Behaviour: Emotion in Invertebrates? Current Biology 21, 765
R463–R465. https://doi.org/10.1016/j.cub.2011.05.028 766
Mesmoudi, S., Rodic, M., Cioli, C., Cointet, J.-P., Yarkoni, T., Burnod, Y., 2015. LinkRbrain: Multi-scale data 767
integrator of the brain. Journal of Neuroscience Methods 241, 44–52. 768
https://doi.org/10.1016/j.jneumeth.2014.12.008 769

26
Molesti, S., Meguerditchian, A., Bourjade, M., 2020. Gestural communication in olive baboons (Papio anubis): 770
repertoire and intentionality. Animal Cognition 23, 19–40. https://doi.org/10.1007/s10071-019-01312-y 771
Mukherjee, D., Lim, W.M., Kumar, S., Donthu, N., 2022. Guidelines for advancing theory and practice through 772
bibliometric research. Journal of Business Research 148, 101–115. 773
https://doi.org/10.1016/j.jbusres.2022.04.042 774
Proust, J., 1999. Intentionality, Consciousness and the System’s Perspective, in: Fisette, D. (Ed.), Consciousness 775
and Intentionality: Models and Modalities of Attribution, The Western Ontario Series in Philosophy of 776
Science. Springer Netherlands, Dordrecht, pp. 51–72. https://doi.org/10.1007/978-94-015-9193-5_3 777
Proust, J., 1995. Intentionality and evolution. Behavioural Processes 35, 275–286. https://doi.org/10.1016/0376-778
6357(95)00057-7 779
Raimbault, B., Cointet, J.-P., Joly, P.-B., 2016. Mapping the Emergence of Synthetic Biology. PLoS ONE 11, 780
e0161522. https://doi.org/10.1371/journal.pone.0161522 781
Schel, A.M., Townsend, S.W., Machanda, Z., Zuberbühler, K., Slocombe, K.E., 2013. Chimpanzee Alarm Call 782
Production Meets Key Criteria for Intentionality. PLoS ONE 8, e76674. 783
https://doi.org/10.1371/journal.pone.0076674 784
Seth, A.K., Baars, B.J., Edelman, D.B., 2005. Criteria for consciousness in humans and other mammals. 785
Consciousness and Cognition 14, 119–139. https://doi.org/10.1016/j.concog.2004.08.006 786
Setiya, K., 2018. Intention, in: Zalta, E.N. (Ed.), The Stanford Encyclopedia of Philosophy. Metaphysics 787
Research Lab, Stanford University. 788
Tancoigne, E., Barbier, M.M., Cointet, J.-P., Richard, G., 2014. Les services écosystémiques dans la littérature 789
scientifique : démarche d’exploration et résultats d’analyse (Research Report). Institut National de la 790
Recherche Agronomique. 791
Tomasello, M., Carpenter, M., 2007. Shared intentionality. Developmental Science10, 121–125. 792
https://doi.org/10.1111/j.1467-7687.2007.00573.x 793
Towner, S., 2010. Concept of mind in non-human primates. Bioscience Horizons 3, 96–104. 794
https://doi.org/10.1093/biohorizons/hzq011 795
Trösch, M., Bertin, E., Calandreau, L., Nowak, R., Lansade, L., 2020. Unwilling or willing but unable: can 796
horses interpret human actions as goal directed? Animal Cognition 23, 1035–1040. 797
https://doi.org/10.1007/s10071-020-01396-x 798
Tuyttens, F.A.M., de Graaf, S., Andreasen, S.N., de Boyer des Roches, A., van Eerdenburg, F.J.C.M., Haskell, 799
M.J., Kirchner, M.K., Mounier, Luc., Kjosevski, M., Bijttebier, J., Lauwers, L., Verbeke, W., Ampe, B., 800
2021. Using Expert Elicitation to Abridge the Welfare Quality® Protocol for Monitoring the Most 801
Adverse Dairy Cattle Welfare Impairments. Frontiers in Veterinary Science 8, 634470. 802
https://doi.org/10.3389/fvets.2021.634470 803
Vasconcelos, M., Hollis, K., Nowbahari, E., Kacelnik, A., 2012. Pro-sociality without empathy. Biology Letters 804
8, 910–912. https://doi.org/10.1098/rsbl.2012.0554 805
Volpato, G.L., 2009. Challenges in assessing fish welfare. ILAR Journal 50, 329–337. 806
https://doi.org/10.1093/ilar.50.4.329 807
Völter, C.J., Lonardo, L., Steinmann, M.G.G.M., Ramos, C.F., Gerwisch, K., Schranz, M.-T., Dobernig, I., 808
Huber, L., 2023. Unwilling or unable? Using three-dimensional tracking to evaluate dogs’ reactions to 809
differing human intentions. Proceedings of the Royal Society B: Biological Sciences 290. 810
https://doi.org/10.1098/rspb.2022.1621 811
Weisz, G., Cambrosio, A., Cointet, J.-P., 2017. Mapping Global Health: A network analysis of a heterogeneous 812
publication domain. BioSocieties 12, 520–542. https://doi.org/10.1057/s41292-017-0053-4 813

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.
Supplementarydata.docx