The world around us is filled with all sorts of information that our brain continuously processes. He receives this information through the sense organs, each of which is responsible for its share of signals: eyes (vision), tongue (taste), nose (smell), skin (touch), vestibular apparatus (balance, position in space and a sense of weight) and ears (sound). By bringing together the signals from all these organs, our brain can build an accurate picture of the environment. But far from all aspects of external signal processing are known to us. One of these secrets is the mechanism of sound source localization.
Scientists from the Laboratory of Neuroengineering of Speech and Hearing (New Jersey Institute of Technology) proposed a new model of the neural process of sound localization. What processes take place in the brain during the perception of sound, how our brain understands the position of the sound source, and how this study can help in the fight against hearing defects. We learn about this from the report of the research group. Go.
Research basis
The information that our brain receives from the sense organs differs from each other both in terms of the source and in terms of its processing. Some signals immediately appear to our brain in the form of accurate information, while others require additional computational processes. Roughly speaking, we feel the touch immediately, but when we hear the sound, we still have to find where it comes from.
The basis for the localization of sounds in the horizontal plane is interaural* time difference (ITD from interaural time difference) sounds reaching the listener's ears.
Interaural base* - the distance between the ears.
There is a specific area in the brain (medial superior olive or MVO) that is responsible for this process. At the moment of receiving a sound signal in the MVO, the transformation of interaural time differences into the reaction rate of neurons takes place. The shape of the velocity curves of the MBO output signal as a function of ITD resembles the shape of the cross-correlation function of the input signals for each ear.
How information is processed and interpreted in MVO is not completely clear, which is why there are several very conflicting theories. The most famous and in fact the classical theory of sound localization is the Jeffress model (Lloyd A Jeffress). It is based on marked line* detector neurons that are sensitive to the binaural synchrony of neural inputs from each ear, with each neuron being maximally sensitive to a certain ITD value (1Π).
Marked line principle* is a hypothesis explaining how different nerves, all of which use the same physiological principles in transmitting impulses along their axons, are able to generate different sensations. Structurally similar nerves can generate different sensory perceptions if they are associated with unique neurons in the central nervous system that are able to decode similar nerve signals in different ways.
Image #1
This model is computationally similar to neural coding based on unlimited cross-correlations of sounds reaching both ears.
There is also a model that suggests that sound localization can be modeled based on differences in the reaction rates of certain populations of neurons from different cerebral hemispheres, i.e. model of interhemispheric asymmetry (1V).
Until now, it has been difficult to unequivocally state which of the two theories (models) is correct, given that each of them predicts different dependences of sound localization on sound intensity.
In the study we are reviewing today, the scientists decided to combine both models in order to understand whether the perception of sounds is based on neural coding or on the difference in the response of individual neuronal populations. Several experiments were conducted in which people aged 18 to 27 years old (5 women and 7 men) took part. Audiometry (measurement of hearing acuity) of the participants was 25 dB or higher at a frequency of 250 to 8000 Hz. The participant of the experiments was placed in a soundproof room, in which special equipment was placed, calibrated with high accuracy. The participants had to, upon hearing the sound signal, indicate the direction from which it comes.
Results of the study
To assess dependency lateralization* of brain activity on the intensity of sound in response to labeled neurons, data on the reaction rate of neurons in the laminar nucleus of the barn owl brain were used.
Laterality* - asymmetry of the left and right halves of the body.
To assess the dependence of the lateralization of brain activity on the reaction rate of certain populations of neurons, we used data on the activity of the lower colliculus of the rhesus monkey brain, after which differences in the speed of neurons from different hemispheres were additionally calculated.
The labeled line model of neuron-detectors suggests that as the sound intensity decreases, the laterality of the perceived source will converge in average values ββsimilar to the ratio of quiet and loud sounds (1S).
The model of interhemispheric asymmetry, in turn, suggests that when the sound intensity decreases to almost threshold levels, the perceived laterality will shift towards the midline (1D).
At higher overall sound intensities, the lateralization is assumed to be intensity invariant (inserts on 1S ΠΈ 1D).
Therefore, the analysis of how the sound intensity affects the perceived direction of the sound, allows you to accurately determine the nature of the processes occurring at that moment - neurons from one common area or neurons from different hemispheres.
Obviously, a person's ability to discriminate between ITDs can vary depending on the intensity of the sound. However, the scientists say it is difficult to interpret previous findings relating sensitivity to ITD and the listener's assessment of the direction of a sound source as a function of sound intensity. Some studies say that when the sound intensity reaches the boundary threshold, the perceived laterality of the source decreases. Other studies suggest that there is no effect of intensity on perception at all.
In other words, scientists are "softly" hinting that there is little information in the literature regarding the relationship of ITD, sound intensity and determining the direction of its source. There are theories that exist as a kind of axioms accepted by the scientific community. Therefore, it was decided to test in detail all theories, models and possible mechanisms of hearing perception in practice.
The first experiment was set up using a psychophysical paradigm, which allowed the study of ITD-based lateralization as a function of sound intensity in a group of ten normal hearing participants.
Image #2
The sound sources have been specifically tuned to cover the majority of the frequency range within which humans are able to recognize ITDs, i.e. 300 to 1200 Hz (2Π).
In each of the trials, the listener had to indicate the estimated laterality, measured as a function of the level of sensation, in the range of ITD values ββfrom 375 to 375 ms. To determine the effect of sound intensity, a non-linear mixed effect model (NMLE) was used that included both fixed and random sound intensity.
Timetable 2V demonstrates estimated lateralization with spectrally flat noise at two sound intensities for a representative listener. A schedule 2S shows raw data (circles) and NMLE-fitted data (lines) of all listeners.
Table No. 1
The table above shows all NLME parameters. It can be seen that the perceived laterality increased with increasing ITD, as expected by the scientists. As the intensity of the sound decreased, perception shifted more and more towards the midline (inset in the graph 2C).
These trends were reinforced by the NLME model, which showed a significant effect of ITD and sound intensity on the maximum degree of laterality, confirming the model of hemispheric differences.
In addition, the average audiometric thresholds of pure tones had little effect on perceived laterality. But the intensity of the sound did not significantly affect the indicators of psychometric functions.
The main goal of the second experiment was to determine how the results obtained in the previous experiment will change when taking into account the spectral features of stimuli (sounds). The need to check spectrally flat noise at low sound intensity is that parts of the spectrum may not be audible and this may affect the direction of the sound. Therefore, the fact that the width of the audible part of the spectrum can decrease with decreasing sound intensity can be mistaken for the results of the first experiment.
Therefore, it was decided to conduct another experiment, but with the application back A-weighted* noise.
A-weighting* applied to sound levels to take into account the relative loudness perceived by the human ear, since the ear is less sensitive to low sound frequencies. A-weighting is implemented by arithmetic adding a table of values ββlisted in octave bands to measured sound pressure levels in dB.
On the chart 2D Raw data (circles) and NMLE-fitted data (lines) of all participants are shown.
Data analysis showed that when all parts of the sound are approximately equally audible (both in the first and second experiments), the perceived laterality and slope on the graph, explaining the change in laterality with ITD, decrease with decreasing sound intensity.
Thus, the results of the second experiment confirmed the results of the first. That is, in practice it was shown that the model proposed back in 1948 by Jeffress is not correct.
It turns out that the localization of sounds worsens with a decrease in sound intensity, and Jeffress believed that sounds are perceived and processed by a person in the same way, regardless of their intensity.
For a more detailed acquaintance with the nuances of the study, I recommend looking at .
Finale
Theoretical assumptions and practical experiments confirming them have shown that brain neurons in mammals are activated at different rates depending on the direction of the sound signal. Next, the brain compares these speeds between all the neurons involved in the process to dynamically build a map of the sound environment.
Jeffresson's model is actually not 100% wrong, as it can be used to perfectly describe the localization of the sound source in fusels. Yes, for barn owls the intensity of the sound does not matter, in any case they will determine the position of its source. However, this model does not work with rhesus monkeys, as previous experiments have shown. Therefore, this Jeffresson model cannot describe the localization of sounds for all living beings.
Experiments with the participation of people once again confirmed that the localization of sounds proceeds differently in different organisms. Many of the participants could not correctly determine the position of the source of sound signals due to the low intensity of the sounds.
Scientists believe that their work shows a certain similarity between how we see and how we hear. Both processes are associated with the speed of neurons in different parts of the brain, as well as with the assessment of this difference to determine both the position of objects we see in space and the position of the source of the sound we hear.
In the future, the researchers are going to conduct a series of experiments to examine in more detail the connection between human hearing and vision, which will allow us to better understand exactly how our brain dynamically builds a map of the world around us.
Thank you for your attention, stay curious and have a great week everyone! π
Thank you for staying with us. Do you like our articles? Want to see more interesting content? Support us by placing an order or recommending to friends, , 30% discount for Habr users on a unique analogue of entry-level servers, which was invented by us for you: (available with RAID1 and RAID10, up to 24 cores and up to 40GB DDR4).
Dell R730xd 2 times cheaper? Only here in the Netherlands! Dell R420 - 2x E5-2430 2.2Ghz 6C 128GB DDR3 2x960GB SSD 1Gbps 100TB - from $99! Read about
Source: habr.com