Toward understanding the communication in sperm whales

Andreas, Jacob; Beguš, Gašper; Bronstein, Michael M.; Diamant, Roee; Delaney, Denley; Gero, Shane; Goldwasser, Shafi; Gruber, David F.; Haas, Sarah; Malkin, Peter; Pavlov, Nikolay; Payne, Roger; Petri, Giovanni; Rus, Daniela; Sharma, Pratyusha; Tchernov, Dan; Tønnesen, Pernille; Torralba, Antonio; Vogt, Daniel M.; Wood, Robert J.

doi:10.1016/j.isci.2022.104393

Cited by 13 publications

(3 citation statements)

References 151 publications

(188 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although we cannot know for certain how whales perceive codas or label the identity of conspecifics, the observed patterns could arise from the selection for identity signals that enable social assortment and cooperation in sperm whales, as has been shown for humans ( 6 , 44 ) and suggested for other taxa ( 5 , 37 , 45 , 46 ). Playback experiments comparing behavioral responses of whales exposed to same-clan, different-sympatric clan, or different-allopatric clan identity codas are essential for validating the patterns described here ( 47 ), but such experimental studies with wild marine mammals remain logistically challenging ( 48 ). In the interim, studies like ours that examine whether specific acoustic signals meet symbolic marker predictions can be informative for cultural species.…”

Section: Discussionmentioning

confidence: 99%

Evidence from sperm whale clans of symbolic marking in non-human cultures

Hersh

Gero

Rendell

et al. 2022

Proc. Natl. Acad. Sci. U.S.A.

Self Cite

View full text Add to dashboard Cite

Culture, a pillar of the remarkable ecological success of humans, is increasingly recognized as a powerful force structuring nonhuman animal populations. A key gap between these two types of culture is quantitative evidence of symbolic markers—seemingly arbitrary traits that function as reliable indicators of cultural group membership to conspecifics. Using acoustic data collected from 23 Pacific Ocean locations, we provide quantitative evidence that certain sperm whale acoustic signals exhibit spatial patterns consistent with a symbolic marker function. Culture segments sperm whale populations into behaviorally distinct clans, which are defined based on dialects of stereotyped click patterns (codas). We classified 23,429 codas into types using contaminated mixture models and hierarchically clustered coda repertoires into seven clans based on similarities in coda usage; then we evaluated whether coda usage varied with geographic distance within clans or with spatial overlap between clans. Similarities in within-clan usage of both “identity codas” (coda types diagnostic of clan identity) and “nonidentity codas” (coda types used by multiple clans) decrease as space between repertoire recording locations increases. However, between-clan similarity in identity, but not nonidentity, coda usage decreases as clan spatial overlap increases. This matches expectations if sympatry is related to a measurable pressure to diversify to make cultural divisions sharper, thereby providing evidence that identity codas function as symbolic markers of clan identity. Our study provides quantitative evidence of arbitrary traits, resembling human ethnic markers, conveying cultural identity outside of humans, and highlights remarkable similarities in the distributions of human ethnolinguistic groups and sperm whale clans.

show abstract

Section: Discussionmentioning

confidence: 99%

Evidence from sperm whale clans of symbolic marking in non-human cultures

Hersh

Gero

Rendell

et al. 2022

Proc. Natl. Acad. Sci. U.S.A.

Self Cite

View full text Add to dashboard Cite

show abstract

“…While biologists have classified the vocalizations into codas, virtually nothing is known of their informational content (Andreas et al, 2022) One outstanding question about codas is whether the acoustic properties have any informational content. In this work, we aim to use the fact that the network used learns to encode informationally meaningful properties in an unsupervised way while being able to generate convincing synthetic whale utterances to discover what might be crucial to the communication system.…”

Section: Why Use Deep Learning?mentioning

confidence: 99%

Approaching an unknown communication system by latent space exploration and causal inference

Beguš¹,

Leban²,

Gero³

2023

Preprint

View full text Add to dashboard Cite

This paper proposes a methodology for discovering meaningful properties in data by exploring the latent space of unsupervised deep generative models. We combine manipulation of individual latent variables to extreme values outside the training range with methods inspired by causal inference into an approach we call causal disentanglement with extreme values (CDEV) and show that this approach yields insights for model interpretability. Using this technique, we can infer what properties of unknown data the model encodes as meaningful. We apply the methodology to test what is meaningful in the communication system of sperm whales (Physeter macrocephalus), one of the most intriguing and understudied animal communication systems. We train a network that has been shown to learn meaningful representations of speech and test whether we can leverage such unsupervised learning to decipher the properties of another vocal communication system for which we have no ground truth. The proposed technique suggests that sperm whales encode information using the number of clicks in a sequence, the regularity of their timing, and audio properties such as the spectral mean and the acoustic regularity of the sequences. Some of these findings are consistent with existing hypotheses, while others are proposed for the first time. We also argue that our models uncover rules that govern the structure of communication units in the sperm whale communication system and apply them while generating innovative data not shown during training. This paper suggests that an interpretation of the outputs of deep neural networks with causal methodology can be a viable strategy for approaching data about which little is known and presents another case of how deep learning can limit the hypothesis space. Finally, the proposed approach combining latent space manipulation and causal inference can be extended to other architectures and arbitrary datasets.

show abstract

“…As conservation and management strategies encompass progressively more extreme solutions, ranging from species de-extinction to translation of non-human communication 1 , novel computational techniques can provide improved methods for processing large quantities of multimodal ecological data. In recent years, advances in machine learning (ML) and deep learning (DL), in particular, have revolutionized the analysis of bioacoustic data, facilitating the development of automated pipelines for processing data and contributing to enhanced conservation tactics across diverse taxa.…”

Section: Introductionmentioning

confidence: 99%

Bioacoustic Event Detection with Self-Supervised Contrastive Learning

Bermant¹,

Brickson²,

Titus

2022

Preprint

View full text Add to dashboard Cite

While deep learning has revolutionized ecological data analysis, existing strategies often rely on supervised learning, which is subject to limitations on real-world applicability. In this paper, we apply self supervised deep learning methods to bioacoustic data to enable unsupervised detection of bioacoustic event boundaries. We propose a convolutional deep neural network that operates on the raw waveform directly and is trained in accordance with the Noise Contrastive Estimation principle, which enables the system to detect spectral changes in the input acoustic stream. The model learns a representation of the input audio sampled at low frequency that encodes information regarding dissimilarity between sequential acoustic windows. During inference, we use a peak finding algorithm to search for regions of high dissimilarity in order to identify temporal boundaries of bioacoustic events. We report results using these techniques to detect sperm whale (Physeter macrocephalus) coda clicks in real-world recordings, and we demonstrate the viability of analyzing the vocalizations of other species (e.g. Bengalese finch syllable segmentation) in addition to other data modalities (e.g. animal behavioral dynamics, embryo development and tracking). We find that the self-supervised deep representation learning-based technique outperforms established threshold-based baseline methods without requiring manual annotation of acoustic datasets. Quantitatively, our approach yields a maximal R-value and F1-score of 0.887 and 0.876, respectively, and an area under the Precision-Recall curve (PR-AUC) of 0.917, while a threshold detector based on signal energy amplitude returns a maximal R-value and F1-score of 0.620 and 0.576, respectively, and a PR-AUC of 0.571. The findings of this paper establish the validity of unsupervised bioacoustic event detection using deep neural networks and self-supervised contrastive learning as an effective alternative to conventional techniques that leverage supervised methods for signal presence indication. Providing a means for highly accurate unsupervised detection, this paper serves as an important step towards developing a fully automated system for real-time acoustic monitoring of bioacoustic signals in real-world acoustic data. All code and data used in this study are available online.

show abstract

Toward understanding the communication in sperm whales

Cited by 13 publications

References 151 publications

Evidence from sperm whale clans of symbolic marking in non-human cultures

Evidence from sperm whale clans of symbolic marking in non-human cultures

Approaching an unknown communication system by latent space exploration and causal inference

Bioacoustic Event Detection with Self-Supervised Contrastive Learning

Contact Info

Product

Resources

About