Investigamos, nesta tese, as bases filogenéticas do léxico humano e o modo como elas foram articuladas às habilidades combinatoriais no curso da evolução. Partimos de uma avaliação da interação entre o léxico e a sintaxe com o intuito de determinar quanto da derivação sintática é dependente de informações lexicais e, em que medida, as propostas disponíveis são coerentes com o desenvolvimento evolutivo humano. Nosso objetivo principal é fornecer uma hipótese para a interface entre o léxico e a sintaxe que seja explicativa e evolutivamente adequada. Para esclarecer as discordâncias empíricas e teóricas observadas, propomos, em contrapartida, uma nova abordagem para a gramática, a Abordagem Isomórfica. Argumentamos que o léxico humano decorre do agrupamento de um conjunto de sistemas pré-adaptados que evoluíram separadamente um sistema conceitual-intencional, um sistema sensório-motor e um sistema funcional, cuja integração é um produto direto da emergência de um sistema combinatorial recursivo. Operacionalmente, buscamos, com essa abordagem, reduzir a influência lexical na formação de um objeto linguístico, na tentativa de assegurar uma isonomia funcional entre o léxico e a sintaxe. Motivamos, adicionalmente, a exaptação de um sistema responsável por derivar as unidades discretas empregadas pelo sistema combinatorial, a que damos o nome de sistema funcional. Compartilhado com primatas não humanos, o sistema funcional justifica-se pelo paralelismo entre a denotação rígida das vocalizações de alerta de primatas não humanos e o conteúdo fixo das unidades funcionais da linguagem humana. Propomos, com base nessa correlação, que os mecanismos cognitivos subjacentes aos sistemas de vocalização primata, instanciados pelo sistema funcional, correspondem aos precursores filogenéticos dos traços formais. Funcionalmente, essa nova abordagem incorpora a visão neoconstrucionista de que a derivação da estrutura sintática independe de informações codificadas nas entradas lexicais. Tal conjectura assegura a autonomia funcional da sintaxe, o que, como resultado, nos aproxima do caminho para se ir além da adequação explicativa. / We investigate, in this dissertation, the phylogenetic bases of human lexicon and how they were articulated with combinatorial abilities in evolution. We begin with an evaluation of the interaction between lexicon and syntax to determine how much of the syntactic derivation is dependent on lexical information, and to which extent the available proposals are consistent with human evolutionary development. Our main goal is to come up with an account for the lexicon-syntax interface that is both explanatory and evolutionarily adequate. In an attempt to eliminate the observed empirical and theoretical divergences, we propose a new approach to grammar, the Isomorphic Approach. We claim the human lexicon arouse from the assemblage of a set of pre-adapted systems that evolved separately viz., a conceptual-intentional, a sensory-motor, and a functional system, whose integration is a by-product of the emergence of a recursive combinatorial system. Operationally, we seek, with this approach, to reduce the burden of lexical influence in the derivation of a linguistic object, with the view to establish a functional isonomy between lexicon and syntax. Furthermore, we motivate the exaptation of a pre-adapted system responsible for deriving the discrete units that feed the combinatorial engine, which we call functional system. Shared with non-human primates, the functional system finds justification in the parallel between the rigid denotation of non-human primate alert calls and the fixed content of human language functional units. We propose, based on this conjecture, that the cognitive mechanisms underlying non-human primate alarm-calling systems, suggestively made available by the functional system, comprise the phylogenetic precursors of human language formal features. Functionally, this new approach incorporates the neo-constructionist view that the derivation of a syntactic structure is independent of lexically encoded instructions. With this premise, we aim to ensure the establishment of an autonomous syntax, which, as a result, brings us closer to the road toward a level of explanation that goes beyond explanatory adequacy.

A habilidade que peixes elétricos possuem de se comunicar por meio de um campo elétrico auto-gerado tem atraído a atenção de diversas áreas do conhecimento por mais de 50 anos. Em particular, peixes elétricos pulsadores emitem um sinal que apresenta diversas similaridades com trens de pulsos de neurônios, tornando-se um modelo animal em neurociência. Com o aumento do poder computacional e com o desenvolvimento de novas ferramentas de aprendizagem de máquina, tornou-se possível investigar interações de dominância entre um par de peixes a nível de cada pulso emitido. Até onde se sabe, a codificação e transmissão de informação se dá por modulações nos intervalos entre pulsos. Assim, a comunicação entre peixes é um problema similar à comunicação entre um par de neurônios em áreas relacionadas do sistema nervoso central: a modulação da taxa de disparo de um neurônio é codificada a partir dos pulsos do outro. Neste trabalho investigamos interações sociais entre pares de Gymnotus carapo, uma espécie altamente territorial. Utilizando análise de séries temporais, técnicas de aprendizagem de máquina e teoria da informação, desenvolvemos uma metodologia para detectar padrões comunicativos nos pulsos emitidos pelos peixes. Além disso, observamos uma relação de causalidade na emissão de padrões: apenas um dos peixes modifica o comportamento futuro de seu coespecífico. A direção desse fluxo de informação parece ligada ao papel de dominância/submissão assumido pelo indivíduo. A partir da literatura sobre fisiologia de emissão de novos pulsos, levantamos novas hipóteses sobre o funcionamento dos sistemas neurais responsáveis pela modulação dos intervalos entre pulsos e sobre como estes sistemas podem ter sua sensibilidade modificada por hormônios secretados durante a disputa por dominância. / Weakly electric fishs ability to communicate through a self-generated electric field has attracted attention from several areas of knowledge for more than 50 years. Particularly, pulse-type electric fish emit signals that exhibits several similarities with neuronal spike trains, becoming a popular animal model in neuroscience. Due to the increase of computational power and the development of new machine learning tools, it is now possible to investigate dominance interactions between a pair of fish at the level of every single pulse. As far as we know, information is coded and transmitted by modulation of interval between pulses. Thus, communication between electric fishes presents several similarities with the communication between neurons from different regions on the central nervous system: the spike rate of one neuron is modulated by the pulses emitted by the other. Here we investigated the social interactions between pairs of Gymnotus carapo, a highly territorial species. Using time series analysis, machine learning techniques, and information theory, we developed a methodology to identify communicative patterns in the pulses emitted by the fish. In addition, we observed a causal relation on the pattern emission: only one of the fish modifies the future behavior of its conspecific. This flow of information seems to be related to the dominance/submission role assumed by each individual. From the literature on the physiology of the emission of new pulses, we developed new hypotheses about the functioning of the neural systems responsible for modulating the intervals between pulses and on how these systems can be modified by hormones secreted during a dominance contest.

This thesis challenges dualistic human and animal ontologies by interpreting science fiction (sf) literature, and argues that whereas words can equivocate and obscure meaning, bodies do not lie. Linguistics and semiology extend the definition of "language" to include human and nonhuman gestures and movement, and posthumanist theory expands definitions of "human" and "animal" to explore species boundaries. Scrutinizing opposing dualisms ultimately questions Western epistemology and authority, allowing for an exploration of embodied animal communications within the larger discourse on species and speciesism. This perspective results in a more comprehensive understanding of the interdependence of all species: human, animal, and "other." Although the fictional texts I employ use fantastic elements to posit hypothetical realities, current scientific research reveals that communication with nonhuman animals is indeed possible. / by Jennifer K. Cox. / Thesis (M.A.)--Florida Atlantic University, 2013. / Includes bibliography. / Mode of access: World Wide Web. / System requirements: Adobe Reader.

Investigamos, nesta tese, as bases filogenéticas do léxico humano e o modo como elas foram articuladas às habilidades combinatoriais no curso da evolução. Partimos de uma avaliação da interação entre o léxico e a sintaxe com o intuito de determinar quanto da derivação sintática é dependente de informações lexicais e, em que medida, as propostas disponíveis são coerentes com o desenvolvimento evolutivo humano. Nosso objetivo principal é fornecer uma hipótese para a interface entre o léxico e a sintaxe que seja explicativa e evolutivamente adequada. Para esclarecer as discordâncias empíricas e teóricas observadas, propomos, em contrapartida, uma nova abordagem para a gramática, a Abordagem Isomórfica. Argumentamos que o léxico humano decorre do agrupamento de um conjunto de sistemas pré-adaptados que evoluíram separadamente um sistema conceitual-intencional, um sistema sensório-motor e um sistema funcional, cuja integração é um produto direto da emergência de um sistema combinatorial recursivo. Operacionalmente, buscamos, com essa abordagem, reduzir a influência lexical na formação de um objeto linguístico, na tentativa de assegurar uma isonomia funcional entre o léxico e a sintaxe. Motivamos, adicionalmente, a exaptação de um sistema responsável por derivar as unidades discretas empregadas pelo sistema combinatorial, a que damos o nome de sistema funcional. Compartilhado com primatas não humanos, o sistema funcional justifica-se pelo paralelismo entre a denotação rígida das vocalizações de alerta de primatas não humanos e o conteúdo fixo das unidades funcionais da linguagem humana. Propomos, com base nessa correlação, que os mecanismos cognitivos subjacentes aos sistemas de vocalização primata, instanciados pelo sistema funcional, correspondem aos precursores filogenéticos dos traços formais. Funcionalmente, essa nova abordagem incorpora a visão neoconstrucionista de que a derivação da estrutura sintática independe de informações codificadas nas entradas lexicais. Tal conjectura assegura a autonomia funcional da sintaxe, o que, como resultado, nos aproxima do caminho para se ir além da adequação explicativa. / We investigate, in this dissertation, the phylogenetic bases of human lexicon and how they were articulated with combinatorial abilities in evolution. We begin with an evaluation of the interaction between lexicon and syntax to determine how much of the syntactic derivation is dependent on lexical information, and to which extent the available proposals are consistent with human evolutionary development. Our main goal is to come up with an account for the lexicon-syntax interface that is both explanatory and evolutionarily adequate. In an attempt to eliminate the observed empirical and theoretical divergences, we propose a new approach to grammar, the Isomorphic Approach. We claim the human lexicon arouse from the assemblage of a set of pre-adapted systems that evolved separately viz., a conceptual-intentional, a sensory-motor, and a functional system, whose integration is a by-product of the emergence of a recursive combinatorial system. Operationally, we seek, with this approach, to reduce the burden of lexical influence in the derivation of a linguistic object, with the view to establish a functional isonomy between lexicon and syntax. Furthermore, we motivate the exaptation of a pre-adapted system responsible for deriving the discrete units that feed the combinatorial engine, which we call functional system. Shared with non-human primates, the functional system finds justification in the parallel between the rigid denotation of non-human primate alert calls and the fixed content of human language functional units. We propose, based on this conjecture, that the cognitive mechanisms underlying non-human primate alarm-calling systems, suggestively made available by the functional system, comprise the phylogenetic precursors of human language formal features. Functionally, this new approach incorporates the neo-constructionist view that the derivation of a syntactic structure is independent of lexically encoded instructions. With this premise, we aim to ensure the establishment of an autonomous syntax, which, as a result, brings us closer to the road toward a level of explanation that goes beyond explanatory adequacy.

This study sought to understand the culture-sharing group of people working within the shelter area of a Pacific Northwest animal shelter through the Ethnography of Communication. About 63% of households in the United States live with a companion animal (Risley-Curtis et al., 2006). Recently, there has been a shift toward closer examination into the ways in which humans interact with animals, particularly companion animals. The guiding questions of this study were: RQ1: What are the cultural communication forms performed in the context of the humane society? RQ2: How do shelter workers communicate about companion animals? RQ3: What cultural meanings are instantiated through communication in this context? This qualitative research approach included 40 hours of participant observation, individual interviews and an analysis of a set of documents and artifacts. Utilizing the Ethnography of Communication components, thematic and pattern analysis, findings revealed use of three main communication forms within the shelter: verbal, written and nonverbal communication and the overarching key theme of relational bonding occurring within an animal-centric organization among 4 relational categories: A. Shelter animals and shelter animals, B. Shelter animals and shelter workers, C. Shelter workers and shelter workers and D. Shelter workers and the public. Processes leading to relational bonding are delineated including detailed speech as well as aspects of "broken bonds" and euthanasia and it's effects within a "no-kill" organization.

Sexual selection is responsible for a great diversity of elaborate male traits. A general female preference for males that have exaggerated traits drives this process, but the reasons females exhibit this preference are often unclear. Recent advances in understanding signal evolution have emerged from studies of receiver psychology that focus on how receivers perceive and process communication signals. I apply the perspective of receiver psychology to understand female preference for elaborate signals in túngara frogs (Physalaemus pustulosus). Male túngara frogs produce advertisement calls of variable complexity. Females exhibit a strong preference for complex to simple calls, but previous studies have not found consistent patterns of preference between calls of variable complexity. In my doctoral research, I investigate the function of variable complexity in túngara frogs. Specifically, I address the following questions: 1) Are calls of variable complexity especially relevant to females in certain contexts? Do males respond to female behavior by increasing their production of complex calls? 2) Does male to female proximity influence female response to call complexity? 3) Are females constrained by their perceptual biology in discriminating differences in call complexity? 4) Can females remember attractive males over silences between bouts of advertising? Is working memory for attractive males dependent upon signal complexity? And 5) Does signal memorability increase with signal complexity in a linear relationship? These studies provide several new perspectives to an understanding of female preference for elaborate signals. Phonotaxis experiments demonstrate that females use elicitation behaviors to influence male production of complex calls, that proximity influences female response to signal elaboration, that females are constrained by their perceptual biology in discriminating between complex calls, that memory can influence the evolution of signal complexity, and that memorability and signal complexity share a non-linear relationship. / text

Ce travail de thèse se présente comme une enquête sur les modalités de l’agir-ensemble interspécifique. L’idée qui a présidé à sa mise en œuvre repose sur la volonté d’élargir les questionnements classiques en sociologie de l’action (comment décrire le vivre-ensemble, quelle forme prend l’ordre social) et en linguistique (comment communique-t-on intelligiblement) à un objet sortant de leurs préoccupations traditionnelles : les interactions sociales entre hommes et chiens. Pour cela, nous analysons des données recueillies lors d’interactions ordinaires et quotidiennes entre chiots en éducation et éducateurs canins, ou entre chiens-guides d’aveugles et personnes non-voyantes.Il s’agit d’un travail empirique de recherche sur les ressources utilisées par hommes et chiens pour agir ensemble et communiquer. Pour cela, nous montrons que les actions communes dans lesquelles ils s’engagent sont réalisées de manière ordonnée, et sont séquentiellement organisées – de sorte qu’elles sont descriptibles avec une certaine systématicité. Cette systématicité, qui exhibe le caractère ordonné des interactions, est traitée comme l’indice d’une forme de socialité qui s’incarne dans l’ajustement mutuel. De ce point de vue, cette thèse se présente également comme un travail théorique sur les formes de la socialité interspécifique. De manière incidente, elle se veut en outre le lieu d’une réflexion épistémologique sur la prise en charge par les sciences humaines et la linguistique d’un objet par tradition réservé aux sciences dites naturelles. / « Non human » is an analytical category that has now entered the realm of sociology. The fact that domestic animals might be agents, and relevant interactants has been evoked and investigated in the most recent literature. The originality of our study does not lie in these arguments. It takes them for granted, and analyzes with systematicity some of the resources used by dogs and their human co-interactants (be they educators or visually impaired persons) to communicate with intelligibility, and make each other’s actions mutually accountable. The study is structured by a leading question: what kind of sociality is at stake between dogs and humans ?The dissertation is divided into two introductory theoretical chapters, and three analytical parts. The first chapter establishes the state of the art, as far as human/animal interaction is concerned. After briefly commenting on the Animal Studies and its opposition to the so-called cartesian position, it ends by introducing the ethnomethodological program as a relevant approach to shed a new light on my object. The second chapter offers an epistemological reflection on the analytical ‘naturalist’ framework worth adopting in order to investigate dog-human sociality. It gives an occasion to discuss the transcription format usually used in CA as an adequate frame to shed light on the sequentiality of actions, as well as on conditional relevance. The three next chapters are grounded on these reflections and are more strictly empirical and analytical. Chapter 3 describes the resources used by dogs and humans to interact with intelligibility and to share perceptive knowledge. It analyzes procedures of shared attention, and mutual orientation (for instance, by mutually orienting toward a relevant object for the ongoing action). Chapter 4 goes further into the analysis of participants’ procedural competencies, and observes the systematicity of sequential formats. Chapter 5 is grounded on these analyses and addresses a “topos” as far as human-animal interaction is concerned: issues of cognition. Drawing on the EM program, it proposes a praxeological approach to cognition that does not focus on dog’s capacities or skills but on the way ordinary practices of practical reasoning are accomplished.The PhD dissertation offers an empirical work on human-animal modalities of living and acting together. It aims at showing that mutual actions participants engage in are orderly accomplished and sequentially organized – and therefore descriptible with systematicity.This systematicity, by exhibiting the orderly character of interactions, is treated as a cue of a form of sociality, embodied in mutual adjustment. In this regard, this thesis offers also some theoretical thoughts on forms of interspecific sociality.At the same time, and more incidentally, it develops epistemological considerations about the reflexive relationships between social sciences, linguistics, and natural sciences in the treatment of this “hybrid” objet.

Orientadores: Eleonore Zulnara Freire Setz, Júlio César Bicca-Marques / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Biologia / Made available in DSpace on 2018-08-22T19:36:22Z (GMT). No. of bitstreams: 1 Caselli_ChristiniBarbosa_D.pdf: 2710988 bytes, checksum: faa0532f7d7f3777fbba617b58a34ae4 (MD5) Previous issue date: 2013 / Resumo: A territorialidade corresponde a uma forma de competição na qual competidores expulsam uns aos outros de áreas contendo recursos, os territórios. A defesa de territórios pode ser flexível, sendo manifestada apenas quando existirem recursos críticos que limitem o crescimento da população (frequentemente representados pela disponibilidade de alimento) e quando estes forem economicamente defensáveis. Os primatas do gênero Callicebus são geralmente descritos como territoriais, embora esta caracterização baseie-se em estudos focados em apenas três das 30 espécies conhecidas desse gênero diverso de primatas Neotropicais. Ainda, a manifestação de comportamentos relacionados à territorialidade se mostrou bastante variável em estudos anteriores, podendo estar relacionada à curta duração desses trabalhos, os quais não contemplaram os efeitos da variação sazonal de recursos alimentares na manifestação desses comportamentos. Nesse contexto, investigamos o comportamento territorial de Callicebus nigrifrons, analisando a influencia da variação na disponibilidade de frutos sobre este comportamento. Buscamos também analisar o papel das vocalizações de longo alcance, geralmente associadas à defesa conjunta de território, mas que também pode estar relacionada a defesa de parceiros reprodutivos. Para isso monitoramos o comportamento de um grupo habituado de C. nigrifrons ao longo de 20 meses (Novembro/2009 - Junho/2011; 730 horas de observação) paralelamente ao monitoramento da disponibilidade de frutos de 22 espécies zoocoricas consumidas por estes primatas. Adicionalmente, analisamos a estrutura dos cantos de longo alcance de nove grupos de C. nigrifrons e relacionamos a estrutura desses cantos aos contextos nos quais estes são utilizados. Também realizamos experimentos com "playbacks" para investigar a importância das vocalizações de longo alcance na defesa conjunta de territórios e de parceiros. As vocalizações de C. nigrifrons apresentaram estrutura hierárquica, cuja combinação de unidades menores, as silabas, dá origem as frases, que são então combinadas em seqüências mais longas. Estas vocalizações de longo alcance são utilizadas tanto para comunicação entre membros do mesmo grupo, como para comunicação entre grupos vizinhos e, embora sejam compostas por unidades vocais semelhantes (mesmas silabas e frases), apresentam estruturas diferentes associadas a estes diferentes contextos comportamentais. Os cantos utilizados para comunicação entre grupos vizinhos foram utilizados principalmente para defesa de recursos alimentares importantes na dieta desses primatas. Diferente de outros Callicebus, C. nigrifrons não exibiu comportamento de defesa e patrulhamento das bordas de sua área de vida, mas focaram o comportamento de defesa em áreas com recursos alimentares importantes e economicamente defensáveis, como arvores frutíferas. Esse comportamento de defesa, tanto pela emissão de vocalizações de longo alcance, quanto pelo uso de outros comportamentos agressivos, resultaram na expulsão de grupos vizinhos que se aproximassem de sua área de uso, e variou em intensidade de acordo com a disponibilidade de frutos no ambiente. A baixa freqüência de encontros entre nosso grupo focal e grupos vizinhos sugere que os comportamentos adotados por C. nigrifrons são efetivos para garantir o acesso prioritário a recursos alimentares importantes em sua dieta / Abstract: Territoriality is a form of competition in which competitors exclude each other from areas containing resources, the territories. Territorial behaviors are flexible and should be adopted only when there is critical resource shortage (which limits population growth) and when these are economically defensible. Callicebus monkeys are usually described as territorial, although this characterization is based on studies focused on three of the 30 known species of this diverse genus of Neotropical primates. Still, the expression of behaviors related to territoriality showed a variable pattern in previous studies. Part of these variations within Callicebus genus could be a result of the short duration of these studies, which did not contemplate the effects of seasonal variation of food resources on territorial behavior. Here we investigated the territorial behavior of Callicebus nigrifrons, evaluating the influence of fruit availability over it. We also evaluated the role of C. nigrifrons loud calls in intra and extra group communication, as well as on food and mate defense. We thus monitored the behavior of a habituated group of C. nigrifrons over a period of 20 months (November/2009 - June/2011; 730 observation hours in total). In parallel to behavioral data collection we monitored the availability of fruits from 22 species eaten by these primates. We also described the structure of loud calls of nine groups of C. nigrifrons and related the structure of these calls to the contexts in which they were used, and performed playback experiments to test the hypotheses of mate and joint territorial defense. Callicebus nigrifrons loud calls are higher hierarchical structures formed by different small units, the syllables, which are assembled to form phrases, which are then assembled to form long sequences, the loud calls. These calls are used in different contexts, involving the communication within and between groups. Although these calls are composed by basically the same syllable and phrase types, the proportional contribution of the different syllable and phrases and the way they are assembled differ between these different contexts. The loud calls used for extra group communication was more clearly associated to joint resource defense of important food recourses in C. nigrifrons diet, the fruits. Unlike previews studies of Callicebus spp., C. nigrifrons did not show the behavior of patrolling and marking range boundaries, but advertised the occupancy of its range via loud call emissions, especially from places close to important and economically defensible food sources, such as fruits. The intensity of the defensive behavior, via loud call emissions and aggressive approach and repulsion of neighboring groups, increased with fruit availability. The low frequency of inter-group agonistic encounter suggest that territories ownership advertisement by C. nigrifrons are effective in maintaining the priority access to important food resources / Doutorado / Ecologia / Doutora em Ecologia

Orientadores: Luiz Octavio Marcondes Machado, Gabriel Corrêa Costa / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Biologia / Made available in DSpace on 2018-08-19T15:46:19Z (GMT). No. of bitstreams: 1 Araujo_CarlosBarrosde_D.pdf: 15759383 bytes, checksum: 3888b2cb944f52bf1ab91fa7eb9cebb4 (MD5) Previous issue date: 2011 / Resumo: O bioma Cerrado é um hotspot de conservação que se encontra em grande risco devido à agricultura ou pastagens. Levando-se em conta que alguns grupos têm maior risco de extinção, e este parece ser o caso dos Psittacidae, e ainda que por muito tempo não havia informações disponíveis sobre essas espécies, a situação era crítica. Isso, no entanto, vem lentamente mudando, e graças ao trabalho de algumas universidades brasileiras já existem informações disponíveis sobre o grupo. Neste trabalho usamos dados adquiridos em campo no Cerrado Central (DF e GO), e também dados presentes na literatura para investigar a alimentação, distribuição de potencial e também a comunicação sonora do grupo. Examinamos a existência de padrões de alimentação entre estas espécies, e também se a dieta pode ser um bom preditor da sua distribuição. Investigamos também a comunicação sonora quanto a presença de chamados específicos para a espécie, dado o grande número de espécies presentes no Cerrado. Também investigamos o alcance de comunicação destas espécies, bem como o impacto que o ruído humano pode ter nesse alcance / Abstract: The Cerrado biome is a conservation hotspot that is currently at great risk, due to agriculture or pasture. On the other hand, some groups have more risk of extinction. That seems to be the case of Psittacidae, as they are one of the most threatened bird groups of the world. For a long time there were no information available on these species, so that the situation was critical. This is slowly changing, and thanks to the work of some Brazilian universities there are now some information available on the group. In this work we use new data acquired at the Central Cerrado, and also data present in literature, to further investigate the feeding, potential distribution and also the sound communication of the group. We examine the existence of feeding patterns among these species, and also if diet may be a good predictor of the species distributions. We also investigate the sound communication. We examine the presence of specific calls for the species, the range of communication of these species, and also the impact that human noise may have in their sound communication / Doutorado / Ecologia / Doutor em Ecologia

Decoding non-human species communication offers a unique chance to explore alternative intelligence forms using machine learning. This master thesis focuses on discreteness and grammar, two of five linguistic areas machine learning can support, and tackles inferring syntax and temporal structures from bioacoustics data annotated with animal behavior. The problem lies in a lack of species-specific linguistic knowledge, time-consuming feature extraction and availability of limited data; additionally, unsupervised clustering struggles to discretize vocalizations continuous to human perception due to unclear parameter tuning to preprocess audio. This thesis investigates unsupervised learning to generalize deciphering syntax and short-range temporal patterns in continuous-type vocalizations, specifically fruit-bats, to address the research questions: How does dimensionality reduction affect unsupervised manifold learning to quantify size and diversity of the animal repertoire? and How do syntax and temporal structure encode contextual information? An experimental strategy is designed to improve effectiveness of unsupervised clustering for quantifying the repertoire and to investigate linguistic properties with classifiers and sequence mining; acoustic segments are collected from a dataset of fruit-bat vocalizations annotated with behavior. The methodology keeps clustering methods constant while varying dimensionality reduction techniques on spectrograms and their latent representations learnt by Autoencoders. Uniform Manifold Approximation and Projection (UMAP) embeds data into a manifold; density-based clusterings are applied to its embeddings and compared with agglomerative-based labels, used as ground-truth proxy to test robustness of models. Vocalizations are encoded into label sequences. Syntactic rules and short-range patterns in sequences are investigated with classifiers (Support Vector Machines, Random Forests); graph-analytics and prefix-suffix trees. Reducing the temporal dimension of Mel-spectrograms outperformed previous clustering baseline (Silhouette score > 0.5, 95% assignment accuracy). UMAP embeddings from sequential autoencoders showed potential advantages over convolutional autoencoders. The study revealed a repertoire between seven and approximately 20 vocal-units characterized by combinatorial patterns: context-classification achieved F1-score > 0.9 also with permuted sequences; repetition characterized vocalizations of isolated pups. Vocal-unit distributions were significantly different (p < 0.05) across contexts; a truncated-power law (alpha < 2) described the distribution of maximal repetitions. This thesis contributed to unsupervised machine learning in bioacoustics for decoding non-human communication, aiding research in language evolution and animal cognition.