Are we sensitive to the difference between a temperate forest and a meadow? Do we finely perceive the sound variations associated with seasonal changes or the difference between dawn and midday? How sensitive are we to the presence of living beings in these environments, to their variety?
All these questions are today addressed in a research program combining cognitive sciences and ecology .
Although the study of "soundscapes" began almost half a century ago with the work of R. Murray Schafer in 1977 and Barry Truax in 1978, our knowledge remains fragmentary as to how the human being, with his ears and his auditory system, perceives the complex acoustic scenes produced by so-called “natural” environments, namely environments marginally affected by human activity.
The ecology of soundscapes
Soundscape ecology or ecoacoustics, a scientific field inspired by the pioneering work of Bernie Krause (1987), is booming today. This discipline using soundscapes at different spatial and temporal scales to explore ecological complexity benefits from the appearance of low-cost autonomous passive recorders (see figure 1).
These allow ecologists and ethologists to record massive, high-quality acoustic databases without interfering with the environment.
Ecoacoustics also benefits from the recent development of sophisticated signal processing techniques, the effectiveness of which is increased by modern machine learning methods.
The establishment of collaborations between ecoacousticians and specialists in human hearing thus opens the possibility of better understanding how we – humans – auditorily perceive natural environments, their composition, the ecological processes at work as well as their variations, that they are of natural origin or induced by human activity.
Four distinct habitats in California
This program begins with a “pilot” study aimed at characterizing the acoustic information conveyed by soundscapes recorded in a California biosphere reserve (see figure 2).
These are four distinct habitats of the same temperate terrestrial biome, that is to say a vast and homogeneous region from a climatic point of view.
The sound data, collected and analyzed by Bernie Krause and his colleagues correspond to high quality recordings made in four seasons (spring, summer, autumn, winter), four periods of the day (dawn, midday, evening, night ) within four sites: a forest, a clearing, a scrub and a meadow.
These numerous sound samples are then transmitted to a computer model of the human auditory system simulating the major stages of sound information processing in the inner ear, brain stem and auditory cortex of a human being. Figure 3 shows the average output of this auditory model.
These representations show how the sound modulation power is distributed as a function of the audio frequency (on the ordinate) and the temporal cadence (on the abscissa) of the signal, for each experimental condition. Figure 3 shows the auditory model output for each habitat, averaged across periods of day and seasons.
These images therefore illustrate the spectral and temporal acoustic information available to a human being. Despite the high acoustic variability of soundscapes, these representations differ greatly between habitats, and suggest that we should be quite capable of discriminating these soundscapes and their variations. To further test this hypothesis, we fed the outputs of this auditory model to classification algorithms (machine learning programs). The results of the simulations are very clear: the classification performances (what is the habitat, time of day, season?) are far superior to the performance due to chance.
Hear the habitat, season and time of day
These initial results provide predictions that can then be tested in humans. A second study, behavioral this time , takes this same base of sound samples and organizes it in order to measure the auditory discrimination abilities of these sounds depending on the habitat, the season and the time of day in hearing adults.
The number of months of exposure to natural soundscapes of these participants – all urban at the time of the experiments – was simultaneously estimated using a questionnaire. These discrimination experiments are carried out using a “forced choice” discrimination protocol aimed at limiting the influence of non-sensory biases, for example a personal preference for a habitat.
During each trial of the experimental procedure, three distinct sounds (one of which comes, for example, from a clearing and the other two from a forest) are presented to the ears of the participants in random order and the latter must determine which of them is these sounds are different from the other two.
The behavioral results presented in Figure 4 are consistent with the model's predictions: we are able to discriminate habitat, season and time of day well above chance. What's more, these abilities vary very little between people tested, and – more surprisingly – these abilities are not correlated with the duration of prior exposure to natural soundscapes. However, human performance is lower than that of the model, suggesting a form of sub-optimality in humans that should be studied.
Listen to the living, the water, the wind or the rain
These first results illustrate basic auditory abilities underlying listening to natural environments.
Other similar work on the nature of acoustic information used by humans to detect the presence of living beings and determine their variety (species richness) is underway thanks to the research programs of F. Apoux, E . Grinfeder, N. Miller-Viacava and R. McWalter within our group.
These studies are motivated by previous results from other teams suggesting that biological sound sources are acoustically distinct from geophysical sources like water, wind or rain , and that our auditory system would perform priority processing of sounds of biological origin. .
Psychological, sensory and emotional effects
Millions of people live in rural areas , and many city dwellers invest time and resources to regularly expose themselves to these natural landscapes within green spaces or national parks. For all these people, access to these biodiverse spaces could contribute to their well-being and quality of life as numerous studies show.
Furthermore, natural soundscapes are gradually and profoundly transformed under the effect of human activities .
Everything suggests that we are sensitive to these transformations and that this will not be without psychological, sensory and emotional effects. Taken together, these data encourage a better understanding of the auditory mechanisms involved in these fundamental interactions between our organism and the ecological processes at work within natural environments.
The research program Auditory perception of natural soundscapes: Hearing biodiversity – HEARBIODIV funded by the National Research Agency (ANR) is supported by the École normale supérieure (Christian Lorenzi), the National Museum of Natural History (Jérome Sueur), Bernie Krause (Wild Sanctuary) and Régis Férrière (IRL CNRS iGlobes).
The National Research Agency (ANR) funds project-based research in France. Its mission is to support and promote the development of fundamental and finalized research in all disciplines, and to strengthen the dialogue between science and society. To find out more, consult the ANR website.
This article is also written as part of the 1st Biennial on Nature and the Living, co-organized by the ENS-PSL, the National Museum of Natural History (MNHN) and the School of Decorative Arts. Find here the program for this event which takes place on September 23.