Doubtful Sound

Points of View

Article by Ann Warde

An invitation to observe and explore the composition Doubtful Sound via tangents, wanderings, diversions, digressions, and detours.

Share this article
go to article

Network Images:
Doubtful Sound 

Doubtful Sound Network
Doubtful Sound Data Visualization 001 300dpi
Screen Shot 2021 07 07 at 3 52 04 PM Cassiopeia Notes
Cacciopo Cassiopeia from WUO Mprogram Booklet cleaned Up Score Only 600dpi

Sounds & Music

Doubtful Sound - Kingston, Ontario

00:00 / 00:00

Field Recordings

These sounds are used in some realisations of Doubtful Sound.
This is a subset of the entire collection of recordings used in performances of Doubtful Sound.
Recorded by Ann Warde.
(Headphones facilitate listening to the many very soft sounds)

Recording Assigned To node B2 (Cassiopeia). Mount St.Angelo, Virginia.

00:00 / 00:00

Recording Assigned To node A6 (Cassiopeia). Edinburgh, Scotland.

00:00 / 00:00

Recording Assigned To node Eb3 (Cassiopeia). Mount St.Angelo, Virginia.

00:00 / 00:00

Recording Assigned To node A2 (Cassiopeia). Brussels, Belgium.

00:00 / 00:00

Recording Assigned To node F#6 (Cassiopeia). Mount St.Angelo, Virginia.

00:00 / 00:00

Doubtful Sound, New Zealand

Doubtful Sound, New Zealand, is a place one can visit, home to a population of bottlenose dolphins. For some history take a look at A Doubtful future, an article in a 2008 issue of the New Zealand Listener.

Data representing observations of the social networks among these animals has been extensively studied. The data contain emergent structures—specifically clustering and coherence: a few individuals reach out to interact with many others, a few individuals receive many interactions from others, information transfer is not disrupted if individual animals disappear from the network.

A non-comprehensive list of information about the data set:

Information and plots of the Doubtful Sound dolphin social network data

Plots and statistics for the Doubtful Sound dolphin social network data

Doubtful Sound dolphin social network data in connection with R statistical software

Animated interactive plot of Doubtful Sound dolphin social network data

Informative articles (with abstracts):

The Bottlenose Dolphin Community of Doubtful Sound Features a Large Proportion of Long-Lasting Associations: Can Geographic Isolation Explain This Unique Trait?

David Lusseau, Karsten Schneider, Oliver J. Boisseau, Patti Haase, Elisabeth Slooten, Steve M. Dawson

Behavioral Ecology and Sociobiology 54, no. 4 (September 2003): 396-405.

ABSTRACT: More than 12 studies of different bottlenose dolphin populations, spanning from tropical to cold temperate waters, have shown that the species typically lives in societies in which relationships among individuals are predominantly fluid. In all cases dolphins lived in small groups characterised by fluid and dynamic interactions and some degree of dispersal from the natal group by both sexes. We describe a small, closed population of bottlenose dolphins living at the southern extreme of the species' range. Individuals live in large, mixed-sex groups in which no permanent emigration/immigration has been observed over the past 7 years. All members within the community are relatively closely associated (average half-weight index>0.4). Both male–male and female–female networks of preferred associates are present, as are long-lasting associations across sexes. The community structure is temporally stable, compared to other bottlenose dolphin populations, and constant companionship seems to be prevalent in the temporal association pattern. Such high degrees of stability are unprecedented in studies of bottlenose dolphins and may be related to the ecological constraints of Doubtful Sound. Fjords are low-productivity systems in which survival may easily require a greater level of co-operation, and hence group stability. These conditions are also present in other cetacean populations forming stable groups. We therefore hypothesise that ecological constraints are important factors shaping social interactions within cetacean societies.

Identifying the role that individual animals play in their social network

David Lusseau, MEJ Newman

Proceedings of the Royal Society of London B, Supplement, 271 (2004): S477–S481

ABSTRACT: Techniques recently developed for the analysis of human social networks are applied to the social network of bottlenose dolphins living in Doubtful Sound, New Zealand. We identify communities and subcommunities within the dolphin population and present evidence that sex- and age-related homophily play a role in the formation of clusters of preferred companionship. We also identify brokers who act as links between subcommunities and who appear to be crucial to the social cohesion of the population as a whole. The network is found to be similar to human social networks in some respects but different in some others such as the level of assortative mixing by degree within the population. This difference elucidates some of the means by which the network formed and evolves.

Animal social networks: an introduction

Jens Krause, David Lusseau, Richard James

Behavioral Ecology and Sociobiology 63 (2009): 967–973

ABSTRACT: Network analysis has a long history in the mathematical and social sciences and the aim of this introduction is to provide a brief overview of the potential that it holds for the study of animal behaviour. One of the most attractive features of the network paradigm is that it provides a single conceptual framework with which we can study the social organisation of animals at all levels (individual, dyad, group, population) and for all types of interaction (aggressive, cooperative, sexual etc.). Graphical tools allow a visual inspection of networks which often helps inspire ideas for testable hypotheses. Network analysis itself provides a multitude of novel statistical tools that can be used to characterise social patterns in animal populations. Among the important insights that networks have facilitated is that indirect social connections matter. Interactions between individuals generate a social environment at the population level which in turn selects for behavioural strategies at the individual level. A social network is often a perfect means by which to represent heterogeneous relationships in a population. Probing the biological drivers for these heterogeneities, often as a function of time, forms the basis of many of the current uses of network analysis in the behavioural sciences. This special issue on social networks brings together a diverse group of practitioners whose study systems range from social insects over reptiles to birds, cetaceans, ungulates and primates in order to illustrate the wide-ranging applications of network analysis.

Are the ‘resident’ dolphins of Doubtful Sound becoming less resident?

Shaun D. Henderson, Stephen M. Dawson, William Rayment, Rohan J. C. Currey

Endangered Species Research 20 (2013): 99–107.

ABSTRACT: Patterns of habitat use by wide-ranging animals may change in response to perturbations within their environment. In 2009, a group of 15 bottlenose dolphins, known to be part of the population in Doubtful Sound, New Zealand, were seen in another nearby fiord. This population has been closely monitored since 1990, and this was the first time that a group from Doubtful Sound was seen elsewhere. Since this first occurrence there have been at least 6 other occasions on which ‘resident’ dolphins were missing for at least 3 d, to later reappear in the fiord. During these absences, the other members of the population were routinely sighted. We use capture-recapture modelling based on photo-ID data to demonstrate a dramatic difference in capture probability between 2005 to 2009 and 2010 to 2011. Given the extremely high capture probability in the first period, and the fact that field effort has significantly increased from 2009 through to 2011, it is unlikely that these groups were within the fiord and simply missed. These findings suggest the possibility that the habitat use of this population has changed to include relatively frequent excursions beyond the fiord complex.

Networks and cultural evolution

One of many uses of network theory is its application to the investigation of how animal social networks might contribute to cultural evolution. Mauricio Cantor and Hal Whitehead’s 2013 paper The interplay between social networks and culture: theoretically and among whales and dolphins is useful:

Culture is increasingly being understood as a driver of mammalian phenotypes [author's note: an organism’s phenotype is the entire set of its observable characteristics]. Defined as group-specific behaviour transmitted by social learning, culture is shaped by social structure. However, culture can itself affect social structure if individuals preferentially interact with others whose behaviour is similar, or cultural symbols are used to mark groups. Using network formalism, this interplay can be depicted by the coevolution of nodes and edges together with the coevolution of network topology and transmission patterns.
Mauricio Cantor and Hal Whitehead . “The interplay between social networks and culture: theoretically and among whales and dolphins.” Philosophical Transactions of the Royal Society B 368 (2013): 20120340.

Cantor and Whitehead’s focus on the co-evolution of nodes and edges on the one hand, and of network topology and transmission patterns on the other, has direct implications for the potential design and organisation of the Doubtful Sound composition project. Specifically, it suggests that the network structure, its topology—a term which envisions the network as a shape in 3-dimensions, may evolve over time. It’s intriguing to consider two network structures which inform the Doubtful Sound composition, and how they may be specifically understood to evolve.

The composition system’s nodes and edges encompass both the pitches and the paths linking them in the Cassiopeia network, and the specific animals and their communication links in the Doubtful Sound network. The network topology and transmission patterns—the current, potentially evolving structure of the network, and how the system travels through it, step-by-step, creating and potentially changing that structure—are engendered in the specifics of each of these networks.

The instructions for the realisation of George Cacioppo’s Cassiopeia specify that: “The performance may involve a single network, or any part of it, all of the networks, or any part of them, with or without fantasy elements.

That is, during a performance, and particularly through a set of performances, only specific aspects of the composition’s overall network structure may form the source of its sounds. And, in a sense, a kind of evolutionary process is free to develop between the networked score and the performer’s realisation of it. A specific performer may come to emphasise certain aspects of the network structure and to overlook or leave aside other aspects. In this way, the flexible organisation of Cassiopeia itself engenders an evolutionary potential with regard to its life and growth as a piece of music.

The Doubtful Sound dolphin network, as it is being used in the Doubtful Sound composition project, takes the initial form of a kind of snapshot of the social interaction behavior of this population of dolphins as they were observed over a specific period of time. An inherent structural principle of this network’s organisation is its ability to maintain a stable functioning in terms of its distribution of information. That is, even if some “nodes” (that is, some individual dolphins) disappear from the network, its basic functional capacity for distributing information to all of its members remains stable. So, the collection of individuals within the network may evolve, but the specific function of distributing information throughout the network will remain.

In terms of the use of the Doubtful Sound network as a contributor to the architecture of the Doubtful Sound composition, this aspect of a stable yet at the same time evolutionary structure becomes a fruitful region for exploration of the evolution of the Doubtful Sound compositional system over time.


Cacioppo, George. “Cassiopeia.” In Notations, by John Cage (New York: Something Else Press, Inc., 1969), 53.

Cacioppo, George. “Cassiopeia.” Donald Bohlen, pianist. On Disk 3:1962-63, Music from the Once Festival 1961-1966, New World Records, 2003.

Cantor, Mauricio and Hal Whitehead. “The interplay between social networks and culture: theoretically and among whales and dolphins.” Philosophical Transactions of the Royal Society B 368 (2013): 20120340.

Coussi-Korbel, Sabine and Dorothy M. Fragaszy. “On the relation between social dynamics and social learning.” Animal Behaviour 50 (1995): 1441-1453.

Hebets, Eileen A. and Alissa Anderson. “Using cross-disciplinary knowledge to facilitate advancements in animal communication and science communication research.” The Company of Biologists Ltd, Journal of Experimental Biology 221 (2018): jeb179978. doi:10.1242/jeb.179978

Henderson, Shaun D., Stephen M. Dawson, William Rayment, Rohan J. C. Currey. “Are the ‘resident’ dolphins of Doubtful Sound becoming less resident?” Endangered Species Research 20 (2013): 99–107.

Krause, Jens, David Lusseau, Richard James. “Animal social networks: an introduction.” Behavioral Ecology and Sociobiology 63 (2009): 967–973.

Laland, Kevin N. and William Hoppitt. “Do animals have culture?” Evolutionary Anthropology 12 (2003): 150–159.

Lusseau, David. “The Emergent Properties of a Dolphin Social Network.” Proceedings: Biological Sciences 270, Supplement: Biology Letters (November 2003): S186-S188

Lusseau, David and MEJ Newman. “Identifying the role that individual animals play in their social network.” Proceedings of the Royal Society of London B, Supplement, 271 (2004): S477–S481.

Lusseau, David, Karsten Schneider, Oliver J. Boisseau, Patti Haase, Elisabeth Slooten, Steve M. Dawson. “The Bottlenose Dolphin Community of Doubtful Sound Features a Large Proportion of Long-Lasting Associations: Can Geographic Isolation Explain This Unique Trait?” Behavioral Ecology and Sociobiology 54, no. 4 (September 2003): 396-405.

Lusseau, David, Hal Whitehead, Shane Gero. “Incorporating uncertainty into the study of animal social networks.” Animal Behaviour 75 (2008):1809e1815.

Marcoux, Marianne and David Lusseau. “Network modularity promotes cooperation.” Journal of Theoretical Biology 324 (2013): 103-108.

Psorakis, Ioannis, Bernhard Voelkl, Colin J. Garroway, et al. “Inferring social structure from temporal data.” Behavioral Ecology and Sociobiology 69 (2015): 857-866.

Rendell, Luke and Hal Whitehead. “Culture in whales and dolphins.” Behavioral and Brain Sciences 24 (2001): 309-382.

Samaniego, Horacio and Melanie E. Moses. “Cities as organisms: Allometric scaling of urban road networks.” Journal of Transport and Land Use 1, no. 1 (Summer 2008): 21–39.

Experiencing a network’s properties through time

A central question in working with networks as a means of structuring time-based behavior (like music and sound art) is how to dynamically realise the perception of potentially intriguing, patterned behavior over time.

What kinds of alternatives are facilitated by directed and undirected network structures?

The relationship between nodes and edges in the Cassiopeia network is “undirected.” That is, a single path leads both away from and toward each node of any pair of linked nodes (i.e. it’s a two-way street, inviting stepping either forward or backward along an edge to arrive at the next node).

[ Network Images: Doubtful Sound & Cassiopeia ]

However, if, for instance, each edge is “weighted” using a value related to incoming amplitude information from an acoustic instrument, which node is next depends on the probability of any specific node being chosen from among a group of nodes. During the initialisation stage of the program, each node in the Cassiopeia network is assigned a specific piece of information (for instance, a specific audio file). Stepping from node to node via the linked edges between pairs of nodes — traversing the Cassiopeia network — results in a sequence of sound files.

The sound files don’t have anything to do with the “network topology”—they are just things found along the path taken among the nodes. However, what we learn about the network while collecting the sound files can include what we find “along the way”—and these findings may be just what allows emergent processes to be perceived. We also learn, in a sense (maybe), details about the network’s potential—about the current, and potentially ongoing, evolution of the network’s topology, and how it might develop and change over time depending on the particular node chosen at any step. Also, because each sound file represents a node, the sequence of sound files defines (and documents) a “transmission pattern”—a specific path through the network.

On the other hand, when we collect information from the Doubtful Sound network’s nodes (either randomly, or as we step from one to the next), we do learn about the network’s structure, since each node is used to store information about itself in terms of how many edges lead to or away from it. This information provides a kind of snapshot of the network’s current structure: it’s what we know about the current network topology. This information could be stored to provide documentation about the history of the network: as that structure perhaps changes (evolves), an overall, ongoing, collection of current structural information would also contain information about how it has been structured in the past.

This is an interesting kind of “experiential” viewpoint, which is determined in the short term: at every node, the possibility exists for the path to take an unanticipated turn. That is, the most likely edge, the one with the highest probability, may nonetheless not be chosen. And, so, we construct a path through the network by undertaking a journey into it, and the sequence of sound files is a kind of documentation of that path (or transmission pattern).

The transmission patterns are represented by an experiential gathering of information node-by-node, deciding at each node just which node will be next. Because they are unpredictable, due to the weights that specify the likelihood that one or another will be chosen, they can evolve and change over time. On the other hand, collecting information about the fixed network topology (potentially fixed in the short term at least) propels the experiencer into a position high above the network, a bird’s-eye-view from which its overall structural organisation is visible.

Network topology (the ways in which nodes and edges are organised), and its relationship to transmission patterns, characterises the kinds of information that are currently generated by the Doubtful Sound network. Like the Cassiopeia network, nodes in this network also contain specific pieces of information. However, rather than containing values assigned by the system, the Doubtful Sound network nodes (in my set of programs) contain information related to properties inherent in each node that are defined by the structure of the network.

The Doubtful Sound network is a “directed” network: its edges lead in only one direction (it’s a “one-way street”). The information associated with each node includes: the number of paths leading to a node (its “in-degree”), the number of paths leading away from it (its “out-degree”), and the probability that one or another path might be chosen (its “edge weight”).

[ Network Images: Doubtful Sound & Cassiopeia ]

This use of the Doubtful Sound network’s in-degree and out-degree values appears to be a kind of “meta” approach to working with the data. That is, while sound files are chosen “experientially” by a process of stepping along paths through Cassiopeia’s network of two-way interconnected nodes, other parameters—including the number of files played simultaneously, the amount of time to wait before files are played, the length of each file to play (as a subset of the whole file), and how loudly each file is played—are chosen by extracting values based on the properties of a specific Doubtful Sound node. Those properties (as has been mentioned) include the number of edges heading toward a node, and the number of edges leading away. Because the structure of the network defines these in-degree and out-degree values, they are fixed for each node. And because they are fixed, they define aspects of its overall “meta-structure”. However, if the relationships between nodes and edges evolve, so that the in-degree and out-degree values for specific nodes change, then this meta-structure itself will evolve.

Perhaps an analogy is useful. Suppose each node is an ornamental fountain in a city of many ornamental fountains, and that these fountains are embedded within a network of one-way streets. (A network of one-way streets is a directed network, because all paths lead either away or toward a node: never both—that would be an undirected network). If you ride your bicycle from fountain to fountain, as you leave each fountain you make a choice, from several one-way streets, of one route to follow to get to another fountain. Of course, pedaling along in-between the fountains you experience, up close, multiple details of a variety of aspects and features of the cityscape. If, then, you decided to stop off and climb the steps up to the top floor of the city’s highest building, you would look out to see, from a birds-eye view, all the fountains at once, with one-way streets heading toward and away from each one.

As time goes on, while riding a bicycle to and from fountain to fountain, you may discover patterns of one-way streets that lead you easily from one to another, and these might become recurring travel patterns for you, although, on some days you may decide to explore unfamiliar routes. Nonetheless, whenever you look out from high above the city, the arrangement of one-way streets and fountains—its structural organisation—remains the same. The properties of this network that you observe (the number of one-way streets heading to and from the fountains) remain fixed—they can be easily seen from above—and in this sense they stand outside time. By experiencing them as travel patterns, on the other hand, it may gradually become clear how these properties structure and organise time: how they function within and over time, and how traversing the network might in fact shape one’s experience of time.

Furthering the story, the two kinds of information provided by the two networks might be: 1) a differently-coloured ticket randomly assigned to each fountain (to perhaps indicate that you had visited it), and 2) GPS coordinates of the fountain marked on a map of the network of fountains (or the number of one-way streets leading to the fountain, or the number leading away). This second kind of information is data related to the structure of the fountain network in some way.

Aspects of the system’s further development include building processes that facilitate the “co-evolution” of nodes and edges, and the “co-evolution” of network topology and transmission patterns.

Some of the aspects of the composition’s sounds that are available for ordering via network data are: the specific field recordings and/or animal voices (in the form of digital audio files), the number of sounds heard simultaneously, the time delay before any specific set of sound files are heard, and the length and amplitude of each sound file along with its initial position and movement within the stereo field.

Amplitude values from an acoustic instrument being played by a performer may be used to “weight” the edges, and the probability that one path might be chosen over another may be based on these edge weight values.

Making choices by stepping from node to node, with decisions about which node is next being made somewhat indeterminately (because of the edge weights), means it is more likely that some pathways through the nodes will be chosen over others. And, if the edge weights do not change, these pathways could become audible as sequential patterns encompassing those nodes approached via the heaviest-weighted edges.

The properties of specific nodes are fixed; randomness arises through how these nodes are chosen. Stepping through the nodes is a somewhat indeterminate process, because of weighting. The weights may be fixed. Or, with interactive amplitude input as the source of edge weight values, the process becomes more indeterminate on a micro level, and may in some way reflect changes in amplitude over time—that is, the process may reflect the shapes (gestures) of amplitude generated by the acoustic instrument(s).

A Composition Project

Doubtful Sound is an ongoing investigation into musical ways in which people might be invited, through listening, to cultivate unanticipated relationships to their surroundings and to each other. Animal vocalisations suggest alternative forms of communication, field recordings draw attention to the environments within which we communicate, and instruments suggest alternatives and may at times provide a human-centredness.

Sounds & Music ]

Each performance of Doubtful Sound is an interim report: an iteration of an evolving system. The project aims to use overlapping network structures to investigate interactivity and the organisation of the playback (and processing) of field recordings and animal voices. Across multiple iterations of the composition, amplitude values from acoustic instruments inform the network’s choice of parameter values to select specific sound files for playback, and/or to variably control amplitude, sound file fragmentation, and spatial location, among other sound properties.

A recently performed instance of the composition employed networks as a means of selecting a sequence of sound files. A pianist equipped with a set of potential responses (mostly sounds made from sources inside a grand piano) was confronted with an unpredictable situation, with almost no information about which file might appear, or when.

[ Sounds & Music ]

I am exploring two sources of networked data:

a. an early graphic score in the form of a network for performance: Cassiopeia (1962), by the experimental composer George Cacioppo. It was premiered by pianist Donald Bohlen.

[ Network Images: Doubtful Sound & Cassiopeia ]

b. animal social network data from the dolphin population that lives in Doubtful Sound, New Zealand, collected by zoologist David Lusseau. Lusseau has published a number of papers about these data from a variety of perspectives. Of specific interest to this project is “The Emergent Properties of a Dolphin Social Network.”

Much of my exploration involves constructing simultaneous, independent realisations of specific paths through network data—stepping from one “node” (e.g. with respect to the Doubtful Sound network data, a dolphin) to another, via the “edge” (path, or connection) between them. Each node in this case represents the number of interactions the dolphin has received (“indegree”) from other animals, and/or the number of interactions it engages in (“outdegree”) with other animals, over a given length of time.

The differences between the numbers of these kinds of interactions experienced by individual animals results in clusters of interactions centered on specific animals. I am interested in how these clusters might engender a structured, emergent sonic texture that is perceptually interesting from a musical perspective. The independent clustering and coherency behaviors within multiple layers of networked data may potentially facilitate perceptions (over time) of distinct, yet similar, related, activity, which may spark observations of simultaneous, independent continuities.

From this viewpoint, a central question for this project is, how might the structural properties of a network be observed and experienced, unpredictably, over time?

[ Experiencing a network’s properties through time ]

Doubtful Sound, New Zealand, is a place we can visit, a place where a dolphin population lives. Data representing observations of the social networks among these animals have been extensively studied. The data set contains emergent structures, specifically clustering and coherence: a few individuals reach out to interact with many others, a few individuals receive many interactions from others, information transfer is not disrupted if individual animals disappear from the network.

On the other hand, we are, as composers, perhaps interested to put ourselves in a “doubtful” relationship to sound, one in which we consider sound from a critical standpoint — peering behind it, around it, underneath it to see what might lie hidden from our initial experiences.

[ Doubtful Sound, New Zealand ]

The composition has both perceptual and aesthetic goals. On the one hand it uses data collected for specific scientific purposes, data that contain structures (coherency, and clustering) whose information about interconnections among dolphins, taken as a whole, can reveal aspects of their social organisation. At the same time, because these data represent a living, social, behavioral, interactive system, the composition based on these data potentially invites fresh encounters with the patterning of sound.

Our perception of the worlds we occupy changes over time, and those changes accompany alterations in our general perception and recognition of patterns. The clustering and coherency behaviors of multiple layers of network-generated sequences can facilitate the perception of distinct, yet similar, layers of activity: multiple, independent continuities. This behavior perhaps characterises (to some extent) our general interest in networks: they help us to explore a kind of changing organism, encompassing simultaneous multiplicities, and containing possibilities for the distinction of coherent, independent, interacting, continuously changing entities. From this point of observation, our gaze may come to rest on its linkage to some of the essentials of the experience of living in the world we currently occupy.

Perhaps the composition itself may in this way be perceived to behave like a kind of organism — living, changing, different — yet also, simultaneously, be understood as to be the same entity each time we encounter it. This simultaneity of change and similarity may provoke an experience and observation of both the passage of time and the variability and change it embodies.

[ Networks and cultural evolution ]

I view the arrangement of network edges and nodes as a means for asking a set of questions. In a sense their structure implies that they are rules governing the network’s behavior. However, from my perspective, they permit a kind of open-endedness characteristic of a variety of answers to a potentially fixed set of questions. The variability of the answers engenders a variability of time: the behavior of events in changing circumstances, that is, of sets of circumstances, each specific to a particular time. This kind of open yet constrained system can result in sequences of events which may stand in a sense outside of time. In turn, it suggests ways in which we may act within the changing and unknown future that always lies ahead. It allows us to, in a sense, try out that future, to see, feel, and experience potentially probable, and improbable, futures.


09 July 2021
Review status
Double-blind peer review

Leave reply

Your browser does not meet the minimum requirements to view this website. The browsers listed below are compatible. If you do not have any of these browsers, click on the icon to download the desired browser.