DETERMINATION OF OPTIMAL STIMULI FOR AN INTERFACE BASED ON EVOICED POTENTIAL

Description

technical field

[0001] This disclosure falls within the domain of the detection of potentials evoked by direct neural interfaces (or BCIs hereafter, for "Brain Computer Interface"). Previous technique

[0002] As an example below, evoked potentials can be auditory, or alternatively, visual, tactile, etc.

[0003] An evoked potential is a signal that appears in electroencephalogram (EEG) signals when a user is exposed to a stimulus. For example, in the SSAEP (Steady-State Auditory Evoked Potential) approach, the stimulus is auditory and consists of a sound with a carrier frequency to which a sinusoidal amplitude modulation is applied. Equipment such as a headset with EEG signal sensors allows the EEG signals to be measured and the modulation frequency to be identified within them.

[0004] An example of a stimulus used for an SSAEP application is a sine wave at a frequency audible to everyone (for example, 1000 Hz) modulated by a second sine wave, for example, at 37 Hz. The 1000 Hz sine wave is called the carrier wave and allows the modulating signal at 37 Hz to be heard. This latter signal constitutes the information of interest.

[0005] Indeed, it is possible to detect stimulation frequencies between approximately 1 and 200 Hz in the auditory areas of the brain. However, within this frequency range of 1 to 200 Hz, the frequencies are low and can be difficult to reproduce and / or hear. Therefore, a carrier frequency is used as a vector, allowing the user to hear these frequencies more easily. A stimulus (a modulating sine wave at 37 Hz) can then generate a measurable auditory evoked potential in the brain at the same frequency of 37 Hz (plus any harmonics). This frequency is measurable in the EEG signals of the person exposed to the stimulus. Thus, the 37 Hz modulation frequency can be detected in the user's EEG signals.

[0006] However, there is no consensus in the scientific community regarding the best stimuli, that is, those capable of triggering the most powerful evoked potentials, and therefore offering the best readability in EEG signals.

[0007] There are several different ways to construct the stimuli, and the recommended frequency ranges vary greatly from one study to another. Part of the explanation likely lies in the fact that EEG signals are highly variable from person to person and vary considerably depending on many parameters (fatigue, context, concentration, etc.). Certain frequencies may therefore be more easily detectable in some people than others. This can also vary in the same person over time and depending on certain conditions (emotional state, external disturbances, fatigue depending on the time of day, etc.). Known systems and processes are described in the document FERNANDEZ-PARRA J MANUEL ET AL: "Review and Application of Auditory Steady State Responses", 2018 IEEE-EMBS CONFERENCE ON BIOMEDICAL ENGINEERING AND SCIENCES (IECBES), IEEE, December 3, 2018 (2018-12-03), pages 661-666, DOI: 10.1109 / IECBES.2018.8626643. Summary

[0008] This disclosure improves the situation.

[0009] To this end, it offers a solution to obtain the best stimuli for each person in a given context.

[0010] According to one aspect, a method is proposed for determining at least one suitable interface signal for a user. This interface is of the type that operates by detecting an evoked potential in a physiological signal of the user in response to an interface signal emitted for the user. The method comprises, following the emission of at least one interface signal, referred to as the "current interface signal": selecting at least the current interface signal as the suitable interface signal for the user when the emission of the current interface signal has caused an intensity of a measured physiological signal of the user exceeding a threshold. The physiological signal resulting from the user's reaction to an emission of the current interface signal comprises a first carrier frequency and a second modulation frequency.

[0011] The aforementioned physiological signal can be, for example, an electroencephalogram signal (or "EEG signal" hereafter).

[0012] The aforementioned interface may be of the BCI type for "Brain Computer Interface" consisting of detecting in the EEG signal collected from a user of this interface a signal called "evoked potential" because it presents the same frequency as the second aforementioned frequency, of modulation, of the stimulus signal presented to the user.

[0013] This process allows for the selection of optimal stimulus signal shapes to maximize the intensity of the generated evoked potentials. Selecting the best stimulation signals for a brain-computer interface (BCI) has the advantage of adapting stimulus properties to different users and usage conditions (emotional state, fatigue, external disturbances). An interface device using these stimuli can therefore be more robust because it is personalized for a given user and their specific usage conditions. Indeed, such brain-computer interfaces (or "BCIs"), which are reactive because evoked potentials are generated in response to stimuli, rely on the analysis of brain activity (by detecting evoked potentials using, for example, a Steady-State Auditory Evoked Potential (SSAEP) method).It is therefore advantageous that the signals causing the strongest evoked potentials are those used primarily by the interface device.

[0014] Thus, it will be understood that the process may involve: emit at least one interface signal, called the "current interface signal", receive the physiological signal resulting from a user's reaction to the emission of the current interface signal, select at least the current interface signal as the appropriate interface signal for the user when the emission of the current interface signal has caused a measured intensity of the user's physiological signal, exceeding a threshold, the current interface signal comprising a first carrier frequency and a second modulation frequency.

[0015] In an implementation, the emitted interface signal may result from a transformation of the current interface signal.

[0016] For example, the process may include: transform the current interface signal until a suitable user interface signal selection is (or could not be) made.

[0017] In a given project, the process may then include: measure the user's physiological signal in response to the emission of the current interface signal, transform the current signal, and repeat the measurement and transformation a plurality of times, in order to retain a plurality of interface signals suitable for the user and whose respective emissions caused an intensity of the user's physiological signal greater than a threshold.

[0018] In a given implementation, this threshold can be a function of an average of the intensities measured successively.

[0019] Thus, in such a design, a statistical analysis of the stimulation signals is proposed in relation to the reaction they elicit in the user.

[0020] In an embodiment, the transformation of the current signal is carried out by modifying at least one frequency among the first and second frequencies mentioned above, and / or by modifying at least one amplitude associated with one frequency among the first and second frequencies mentioned above, relative to an amplitude associated with the other frequency among the first and second frequencies.

[0021] These modifications can be made in successive steps until satisfactory frequency ranges are reached, for example, for a given individual.

[0022] Alternatively, however, the transformation of the current signal is carried out by adding random noise to a time-frequency representation of the current signal.

[0023] Furthermore, a retainable interface signal may consist of an initial noisy portion followed in time by at least a second, noise-free portion, called the "pause time." This type of signal can induce evoked potentials with a high signal-to-noise ratio.

[0024] In one implementation, the interface signal suitable for the user is retained by the implementation of inverse correlation filters.

[0025] The method known as "inverse correlation filters" is a psychophysical method commonly used to study cognitive processes without prior knowledge of the impact of a particular stimulus on an individual's cognitive system. Subjects are exposed to such stimuli, and their responses are observed through statistical analysis of the responses associated with each stimulus in order to deduce properties of the underlying processes. This allows for the optimization of stimuli based on the collected responses.

[0026] For example, if the random transformation of an original signal distorts its modulation frequency, shifting it from, say, 35 Hz to 37 Hz, and this transformed signal elicits a strong evoked potential, then this 37 Hz modulation frequency could be used for a new generation of signals to be tested on the same individual. It can thus be understood that randomly added noise can improve interface signals.

[0027] In an embodiment where the interface signals selected are sound signals, the method may further include data storage of the signals selected as suitable for the user (for example the first and second frequencies and associated amplitudes) for the purpose of sound reproduction of several of these signals by loudspeakers, simultaneously.

[0028] In such a design, the user can simultaneously listen to several audio signals at different modulation frequencies. For example, these signals could be played through separate speakers, spaced apart to allow the user to focus on one signal in particular. By focusing on one of these signals, their brain generates a periodic wave with the same frequency as the modulation frequency of the signal they focused on. An evoked potential is thus generated for that signal. If this signal (which generates an evoked potential) is associated with a specific command (for example, turning up the volume on a television), then the detection of the corresponding evoked potential will generate a corresponding command (such as turning up the volume on that television).If the user had focused on another sound signal (of a different modulation frequency for example), another command would have been executed (for example "change television channel", or something else).

[0029] A suitable carrier frequency for an audio signal from such a BCI interface is, for example, between 500 and 3500 Hz. The modulation frequency, on the other hand, can be between 1 and 100 Hz, and more specifically between 20 and 80 Hz.

[0030] These are starting ranges from which the frequencies of initial interface signals can be selected, which can then be tested and transformed to modify, for example, their carrier frequency and / or their modulation frequency.

[0031] With regard to the evaluation of the intensity of a user's physiological signal, the measurement of this physiological signal for a user, in reaction to a current interface signal emission, may involve a search for a frequency in the physiological signal corresponding to the second frequency, of modulation of the interface signal.

[0032] More specifically, the user's "physiological signal intensity" can be determined by estimating the signal-to-noise ratio of the physiological signal measured on the user.

[0033] Thus, in a given implementation, the process may include: retain at least the interface signal suitable for the user, the emission of which caused a maximization, at the modulation frequency, of the signal-to-noise ratio of the physiological signal measured on the user.

[0034] In another aspect, a computer program is proposed that includes instructions for implementing all or part of a process as defined herein when executed by a processor. In another aspect, a non-transient, computer-readable recording medium is proposed on which such a program is recorded.

[0035] According to another aspect, a device is proposed comprising a processing circuit for the implementation of the above process. Brief description of the drawings

[0036] Other features, details, and advantages will become apparent upon reading the detailed description below and analyzing the attached drawings, on which: Fig. 1 [ Fig. 1 ] shows an example of applying the above process to a BCI type interface according to one embodiment. Fig. 2 [ Fig. 2] shows an example of a signal generated with a carrier frequency fp and a modulation frequency fm (above) and (below) an evoked potential signal detected in a user with a frequency identical to that of the modulation frequency fm. Fig. 3 [ Fig. 3 ] shows an example of a process within the meaning of this description, according to one embodiment. Fig. 4 [ Fig. 4 ] shows an example of a device as defined in this description, according to one embodiment. Description of the implementation methods

[0037] We refer first to the figure 1To reiterate the principle of a BCI (Brain-Computer Interface), a UT user wears a headset equipped with electroencephalogram (EEG) signal sensors. This headset can be part of a general BCI interface. Various stimuli (auditory, visual, or other), each with its own modulation frequency, are presented to the user, who then focuses on one of the stimuli. The same modulation frequency as the stimulus on which the user focused can then be measured in their EEG signals. In the example of a stimulus delivered by a light source, the light might flash at the aforementioned modulation frequency. In the case of an auditory signal, the signal might have sinusoidal modulation or consist of successive beeps at the aforementioned modulation frequency.An analysis of the user's EEG signals reveals the presence of a wave frequency corresponding to the aforementioned modulation frequency. Therefore, it is possible to design a BCI interface, according to which several stimuli from respective sources HP1, HP2, HP3, having different modulation frequencies and representing, for example, different respective instructions for operating a machine (e.g., "turn left", "turn right", "brake"), are presented simultaneously to a user, and the user focuses on one of them so that their EEG signals reveal the frequency of one of the HP1 stimuli, and the function associated with this stimulus is then executed by the machine.

[0038] In the example of the figure 1The stimuli are delivered by respective loudspeakers HP1, HP2, HP3... and the user focuses on one of these sound sources. Thus, we describe below an implementation in which the interface signal (stimulus applied to the user UT) is, for example, an audio signal to detect a physiological signal of the SSAEP type (for "Steady-State Auditory Evoked Potential"). The stimulus is therefore auditory and consists of a sound at a carrier frequency fp to which a sinusoidal or square wave modulation is applied, at a modulation frequency fm. The stimuli are generally constructed with a carrier frequency (corresponding to the period Tp of the figure 2 ), and modulated in amplitude, power or energy, by a signal having a second frequency (corresponding to the period Tm of the figure 2 If the stimulation signal is effective, the EEG signal measured by the BCI interface headset has a frequency corresponding to the period Tm of the figure 2 .

[0039] In the case of a visual stimulus, a light source can have a particular color, corresponding to a wavelength LO associated with a carrier frequency fp by a relationship of the type LO = c / fp, where c is the speed of light. Some users may be more sensitive to certain colors than others to maintain their attention. Thus, the choice of color (and therefore of the carrier frequency fp) can be important, especially at certain times of day for the same user, to efficiently collect EEG signals revealing a usable evoked potential (i.e., having a frequency corresponding to the modulation or "flickering" frequency of the light source).For example, a wavelength of 0.5µm can be used for blue as the color of visual stimuli in some subjects, or a wavelength of 0.55µm for green to which some users are more sensitive, or even 0.65µm for red, a color to which other users are even more sensitive, etc.

[0040] It is thus understood that a subjectivity which is personal to each user means that certain stimuli are more easily perceived and can be the source of evoked potentials exploitable to design a robust BCI interface for a given user.

[0041] The following are presented with reference to the figure 3A method for selecting the most effective stimuli for a given user. In step S1, a carrier frequency signal fp with frequency modulation fm is generated and applied as a stimulus to the user. In step S2, the user's EEG signal is recorded, and the aim is to detect a frequency corresponding to the modulation frequency fm within this signal S(EEG). Stimulus signals that successfully generate an evoked potential in step S2, for example, are stored in memory in step S3. From these stimulus signals, successive versions of these signals can be progressively transformed in step S4. by proceeding, for example, by successively modifying the modulation frequency fm and / or the carrier frequency fp, and / or their respective associated amplitudes to construct the stimulus signal SHP, according to a first embodiment, and / or by applying random variations in time, of noise, to a time-frequency representation of the stimulus signals stored in memory at step S3, according to a second embodiment, then by implementing an "inverse correlation filtering" technique, described in detail later.

[0042] Next, the transformations of stimuli that caused evoked potentials are in turn stored in memory (looping back through steps S1 to S3 of the figure 1In step S5, transformed versions of stimuli that elicited evoked potentials with a signal-to-noise ratio (EEG signal at the modulation frequency fm) above a threshold HRR (Hib-Related Ratio) can be retained. This HRR can be fixed or estimated based on the average signal-to-noise ratio. For example, all stimuli with a signal-to-noise ratio above the average signal-to-noise ratio (possibly plus one or more standard deviations) can be retained. In step S6, if some modulation frequencies fm and / or carrier frequencies fp appear whose stimulus signals yielded particularly satisfactory results in test S5, stimulus signals can be constructed with averaged fm and / or fp frequencies from these "satisfactory frequencies" to retest new stimulus versions based on these average frequencies (looping through steps S1 to S5).In particular, when adding time-varying noise, it is possible to create modulation by adding this noise, which has the effect of randomly modifying the modulation frequency, or even the carrier frequency. In this case, the new versions tested may have these modulation and / or carrier frequencies modified. Finally, the best stimulus signals are stored in memory at stage S7 (fm and fp frequencies, associated amplitudes, possibly a form of random noise modifying the signal, etc.). These "satisfactory" signals stored at stage S7 can thus correspond to effective stimuli for generating evoked potentials.

[0043] It is also possible to store signals with noisy parts at the S7 stage, possibly with "pause times" between the noisy parts which correspond to noise-free signal.

[0044] In step S7, the tests performed in steps S1 to S6 on a given user UT (User Technology) yield satisfactory stimulus signals for a BCI (Broadband Control Interface) that this user would use. However, certain general trends have been observed, at least among populations of individuals, such that the signals specific to a user UT could be more broadly suitable for an entire category of users. Thus, in step S8, it is possible to average "satisfactory" signals based on tests performed on several different users, but for example, those belonging to the same category (such as an age group, a professional activity, or other). Therefore, it is possible to further refine the individualization of stimulus signals specific to a user, once their category has been identified in step S8, by repeating the signal optimization steps (as illustrated by the dashed arrow from step S8). figure 3 Thus, it should be noted that at the end of step S7, the signals obtained for a user can be used for that user's BCI interface, and, to go further (in the implementation shown in dotted lines of step S8 of the figure 3 ), these signals tested on different users can be averaged to be standard signals from which individual optimizations can be pursued by each user (once their category has been identified).

[0045] For audio signals, for example, it has been observed that a carrier frequency range (fp) between 500 and 3500 Hz shows promise for rapidly obtaining evoked potentials in a large number of users. The modulation frequency (fm) is generally lower, for example, in the range of 1 to 100 Hz. It has been observed, in particular, that between 20 and 80 Hz, evoked potentials can be detected in a large number of users, and younger users may be more sensitive to higher frequencies (around 60 Hz up to 80 Hz).

[0046] The carrier signal can be modulated at this lower frequency, fm, by a sine wave or a square wave. It is also possible to introduce silences into the carrier signal to generate a signal-silence pattern at a frequency fm corresponding to that of the aforementioned modulation. In step S4, it is also possible to modify the amplitude of the carrier signal, just like the modulation amplitude.

[0047] Auditory stimuli are generated successively with a modification of one or more of these parameters (frequency / amplitude; carrier / modulation; sine / square wave function). Evoked potentials that may occur in response to these stimuli are detected in the user's EEG signals. The aforementioned parameters of amplitude and frequency can be stored in memory at step S3, corresponding to a measurement of the signal-to-noise ratio of the EEG signals at that modulation frequency.

[0048] These parameters can be varied in incremental steps (for example, increasing the modulation frequency step by step from 20 Hz to identify a few modulation frequencies that are particularly suitable for the user because the evoked potentials detected exhibit a high signal-to-noise ratio at these frequencies). Alternatively, these parameters can be varied randomly or according to a predefined probability distribution.

[0049] In particular, one possible implementation employs an "inverse correlation method." This technique relies on the subjectivity of the tests. The transformation of the stimulus presented to the user is random, and the user's reaction may be unexpected. For example, starting with a modulation frequency of 30 Hz, the user may respond less well at 35 Hz than at 30 Hz, but better at 40 Hz than at 30 Hz, and this can be reproduced. This can be explained by cognitive processes specific to the user.

[0050] Based on this observation, a time-frequency transformation is applied to the stimulus signals, and random noise (varying randomly over time) is added to this time-frequency representation to generate new stimuli to be presented to the user. Such a transformation by adding random noise can modify, for example, the carrier frequency and / or the modulation. If the user reacts with a strong evoked potential (measured by their signal-to-noise ratio in the S5 test), then the signal characteristics of such a stimulus are stored in memory at step S7 to become part of the interface signals suitable for the user.

[0051] Thus, instead of finely varying the identified parameters (carrier frequency, modulating signal frequency, carrier amplitude, modulating signal amplitude, etc.), as described in the first embodiment, it is proposed, in this second embodiment, to add noise to the stimuli at step S4 in order to create at step S6 a new set of stimuli to evaluate (loop back to step S1 again).

[0052] Each UT subject listens to these signals. Their EEG signals are recorded synchronously. It is then possible to determine which signals triggered the strongest evoked potentials. The optimal stimuli for a given user can then be identified.

[0053] In one experiment, the aim is to construct the best stimuli using the so-called "inverse correlation" approach. Inverse correlation filtering is a psychophysical method for studying cognitive processes. For example, in other techniques, a highly noisy image of a face can elicit different emotions in different subjects (e.g., fear for some individuals because the face is perceived as angry, or joy because the face is perceived by others as smiling). Inverse correlation thus utilizes the user's highly personal response to a noise-based stimulus. The cognitive system is treated here as a black box. Subjects are exposed to these stimuli, and their responses are observed. Statistical analysis of the responses associated with each stimulus allows researchers to deduce the properties of the underlying processes.Thus, it is possible to optimize stimuli based on the collected responses. Initially, several stimuli are generated from a given stimulus. The given stimulus and the stimuli constituting variants of the given stimulus form a set of stimuli to be tested. Then, at least some of the stimuli from this set of stimuli to be tested are reproduced successively.

[0054] Thus, audio stimuli are considered here as an image by transforming them into their time-frequency representation. Their spectrogram representation can be obtained by dividing the signal into frames and transforming each frame into a frequency representation, to obtain a time-frequency image representing each stimulus.

[0055] Next, for each reproduced stimulus, an evoked potential is detected, and a statistical analysis is performed, including inverse correlation between the detected evoked potential and the reproduced stimulus that triggered it. Specifically, for each detected evoked potential, the power (or SNR, as seen previously) of the evoked potential signal is measured.

[0056] Stimuli are collected / selected from the reproduced stimuli based on the result of the statistical analysis, in particular based on the signal intensity of the detected evoked potentials (the stimuli corresponding, for example, to the detected evoked potentials whose signals are the most intense or to the detected evoked potentials whose intensity is greater than a given power threshold).

[0057] Optionally, the collected stimuli can be averaged, as described previously. In particular, in order not only to optimize the stimuli but also to personalize them, selection and / or averaging are performed based on evoked potentials detected for a given user.

[0058] In practice, several stimuli with different carriers and modulation frequencies can be used. Each stimulus thus obtained is then transformed into an image using a time-frequency representation (spectrogram). Each spectrogram is perturbed by random noise (the image is noisy). The resulting noisy spectrogram is then transformed back into an audio signal.

[0059] These disturbed audio stimuli can be used again, possibly multiple times, to feed a subjective test based on the principle of inverse correlation, as described above. Different subjects can listen to these signals. Their EEG signals are recorded synchronously. It is then possible to determine which signals triggered the strongest evoked potentials. By "strongest," we are referring to the signal-to-noise ratio (SNR) of the EEG signal. The aim is then to determine the stimuli that maximize the SNR at the frequencies measured in the recorded EEG signals, particularly at the modulation frequency. For example, an SNR threshold can be defined above which evoked potentials are considered robust and valid for detection. All stimuli meeting this criterion are collected.The waveforms of these stimuli can then be averaged individually for each subject in order to obtain optimal personalized (i.e. individual) stimuli for a given subject.

[0060] Thus, the above technique offers the following advantages: Automatic selection of stimulus properties giving the best performance, particularly in terms of detectability and response time, Automatic adaptation of a BCI interface to different users, Robustness of detection to changes in usage conditions (emotional state, fatigue, external disturbances).

[0061] We illustrated on the figure 4 A possible implementation of a device to carry out the above technique. The device may typically include: an INT2 interface to emit SHPi stimulation signals intended, for example, to power loudspeakers (or light sources), another INT1 interface to receive EEG signals from the BCI headset sensors, a PROC processor to perform the tests of the figure 3 specifically, and identify the SHPi signals causing the most intense evoked potentials in the S(EEG) signals, and a MEM memory storing the data of these SHPi signals for the implementation of step S7 of the figure 3 (or from step S8 if an averaging is performed over several individuals).

[0062] In particular, MEM memory can be a memory block that also stores instruction data for a computer program, with the PROC processor cooperating with the MEM memory to read this data and execute these instructions.

[0063] The technique described above can find many applications using "reactive" BCI type interfaces (reaction to stimuli), for paralyzed or blind people, or for "thought control" using a BCI interface headset (particularly in the context of connected home control with commands such as "turn on the television", "change channel", "turn on the light", "turn off the heating", etc.).

[0064] It should be noted that the figure 4 can also represent the hardware elements of such a BCI interface. For example, the HPi speakers of the BCI interface can correspond to the earpieces of a stereophonic headset, and at least some of the signals held in memory at step S7 can be played simultaneously by the earpieces of the headset at their respective stereo "positions".

Claims

1. Method for determining at least one interface (BCI) signal appropriate to a user, said interface being of the type that operates by detecting a potential produced in a physiological signal (EEG) of the user (UT) in response to the output of an interface signal to the user, the method comprising, following output of at least one interface signal, known as the current interface signal: selecting at least the current interface signal as the interface signal appropriate to the user when output of the current interface signal has caused the intensity of a measured physiological signal (EEG) of the user to exceed a threshold, the physiological signal resulting from a response of the user to output of the current interface signal, comprising a first frequency, carrier frequency, and a second frequency, modulation frequency.

2. Method according to Claim 1, wherein the interface signal that is output results from a transformation of the current interface signal.

3. Method according to either one of the preceding claims, comprising: - transforming the current interface signal as long as no interface signal appropriate to the user is selected.

4. Method according to any one of the preceding claims, comprising: - measuring the physiological signal (EEG) of the user in response to output of the current interface signal, - transforming the current signal, and - repeating the measurement and transformation a plurality of times in order to retain a plurality of interface signals appropriate to the user that, when output, have each caused the intensity of the physiological signal (EEG) of the user to exceed a threshold.

5. Method according to Claim 4, wherein said threshold is dependent on an average of the successively measured intensities.

6. Method according to one of Claims 2 to 5, wherein the current interface signal is transformed by changing at least one of the first and second frequencies.

7. Method according to one of Claims 2 to 6, wherein the current signal is transformed by changing at least one amplitude associated with one of the first and second frequencies relative to an amplitude associated with the other of the first and second frequencies.

8. Method according to one of Claims 2 to 7, wherein the current signal is transformed by adding random noise to a time-frequency representation of the current signal.

9. Method according to one of the preceding claims, wherein the retained signal comprises at least a first, noisy portion followed in time by a second, non-noisy portion.

10. Method according to one of the preceding claims, wherein said at least one interface signal appropriate to the user is retained by using inverse correlation filters.

11. Method according to any one of the preceding claims, wherein the interface signals are sound signals, and the method further comprises storing data of the retained signals as appropriate to the user for audio reproduction, by loudspeakers, of at least some of said retained signals simultaneously.

12. Method according to one of the preceding claims, comprising measurement of the physiological signal (EEG) of the user in response to the output of a current interface signal by looking for a frequency in the physiological signal (EEG) that matches the second frequency.

13. Method according to one of the preceding claims, wherein the intensity of the physiological signal (EEG) of the user is determined by estimating a signal-to-noise ratio of the physiological signal (EEG) measured on the user.

14. Method according to Claims 12 and 13 taken in combination, comprising: - retaining at least the interface signal appropriate to the user that, when output, has caused maximization, at the modulation frequency, of the signal-to-noise ratio of the physiological signal (EEG) measured on the user.

15. Computer program comprising instructions for carrying out the method according to one of the preceding claims when this program is executed by a processing circuit.

16. Device comprising a processing circuit for carrying out the method according to one of Claims 1 to 14.