JP2010036026A - Method and apparatus for determining human's preference based on brain information - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and an apparatus for measuring information under unconsciousness which give no physical/mental stress on a subject while suppressing introduction cost and maintenance cost. <P>SOLUTION: The method is used for constructing a prediction model of preference by using brain activity information measured at a specified area. The specified area is one of an auxiliary part B which has a vertical width of 10% from the bottom to 30% from the top of a distance between Nz and Cz when the direction from Fpz to Cz according to the ten-twenty electrode system of International Federation is set as the upper side and a horizontal width of almost 4% each of the distance between T3 and T4 from Fpz in the directions of T3 and T4, an auxiliary part C1 which has a vertical width of 10% of the distance between Nz and Cz from the bottom when the direction from Fpz to Cz is set as the upper side and a horizontal width of 10% each of the distance between T3 and T4 from Fpz in the directions of T3 and T4, and an auxiliary part C2 which has a vertical width of 7.5% to 20% of the distance between Nz and Cz in the direction from Fpz to Cz and a horizontal width of 7.5% to almost 20% of the distance between T3 and T4 in the direction from Fpz to T3. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a method and apparatus for discriminating human preferences from brain information.

  Traditionally, human preferences have been obtained primarily through questionnaires. However, in a questionnaire survey, it was impossible to quantitatively measure unconscious and emotional preferences.

  By the way, in recent years, studies have been conducted to measure information in the human unconscious using brain information. For example, Patent Document 1 describes that the user's attention area is measured from the user's brain information using NIRS (Near InfraRed Spectroscopy) to control the operation of the electric wheelchair.

JP 2007-20882 A

  However, since these studies attempt to measure information in the human unconsciousness after measuring the entire brain, there is a problem that a large-scale facility is required, and introduction costs and maintenance costs become very high. . In addition, because the measurement environment is fixed and body movement restrictions are severe, physical and mental stress is given to the subjects. For this reason, it is difficult to apply to other than research, that is, application to daily life, and it has been put to practical use.

  The present invention has been made in view of the above points, and provides a method and apparatus that performs measurement of unconscious information while suppressing introduction cost and maintenance cost, and does not apply physical or mental stress to a subject. The purpose is to do.

  (1) A method for constructing a prediction model of a user's preference relationship with respect to presented information, the step of measuring brain activity information of a specific part by a non-invasive brain activity measurement unit, and the measured brain And constructing a prediction model of the preference relationship from activity information, and the specific region is defined as a position between the nose root Nz and the parietal portion Cz with the top Cz direction as the top from the origin Fpz in the international 10-20 method. Vertical axis width from 10% to upper 30%, and approximately 4% between the left anterior pinna point T3 and the right pinna point T4 in the direction from the origin Fpz to the left pinna point T3 and right pinna point T4, respectively. An auxiliary part B having a horizontal axis width of each, a vertical axis width of 10% below the nose root part Nz and the parietal part Cz from the origin Fpz to the parietal part Cz direction, and the left anterior pinna point T3 from the origin Fpz and Right in front of the left pinna in the direction of the right pinna point T4 Auxiliary part C1 having a horizontal axis width of 10% between T3 and right anterior point T4, or 7.5% to 20% between nasal root Nz and parietal Cz in the direction of parietal Cz from origin Fpz Auxiliary region C2 having a vertical axis width and a horizontal axis width of 7.5% to approximately 20% between the left anterior pinna point T3 and the right pinna point T4 in the direction from the origin Fpz to the left pinna point T3. Any one of the methods described above.

  According to the method (1), since the brain activity information is used when constructing the prediction model of the user's preference relationship, it is possible to construct a prediction model that reflects unconscious preference and emotional preference. it can.

  In addition, since the brain activity information is measured from a specific part called the auxiliary part B, the auxiliary part C1, or the auxiliary part C2, there is no need to measure the entire brain, and the brain activity measuring unit can be downsized. As a result, it is possible to construct a prediction model for user preferences at relatively low introduction costs and maintenance costs. In addition, since no physical movement restriction is imposed on the user, physical and mental stress is not applied to the user.

  In particular, since these specific parts are suitable for prediction of preference relationships, a highly accurate preference relationship prediction model can be constructed.

  (2) A method of presenting additional information about information preferred by a user using the prediction model constructed by the method according to (1), the step of presenting specific information to the user, and the brain The user likes the presented information from the step of measuring the brain activity information for determination of the specific part when the specific information is presented by the activity measuring unit, and the brain activity information for determination and the prediction model And a step of presenting additional information about the information on the condition that it is determined to be preferred.

  According to the method (2), since it is possible to present additional information about information preferred by the user while reducing the size of the brain activity measurement unit, for example, application of information exchange via the Internet or the like is possible. It is possible and can be made practical.

  (3) The method according to (1) or (2), wherein the specific portion is 20% of the vertical axis width between the nose root Nz and the parietal portion Cz from the origin Fpz to the parietal portion Cz in the international 10-20 method. And a main part A having a horizontal axis width of approximately 4% between the left anterior pinna point T3 and the right anterior pinna point T4 in the direction from the origin Fpz to the left pinna point T3 and the right pinna point T4, respectively. A method characterized by:

  According to the method (3), it is possible to improve the accuracy with respect to the prediction (determination) of the preference relationship.

  (4) A user terminal that is communicably connected to a server terminal capable of constructing a prediction model for predicting a user's preference relationship from the user's brain activity information, and has no invasiveness to measure brain activity information of a specific part A brain activity measuring unit; and a communication unit that transmits the brain activity information measured by the brain activity measuring unit to the server terminal, and the specific part is moved from the origin Fpz to the parietal part Cz in the international 10-20 method. The vertical axis width from the lower 10% to the upper 30% between the nasal root Nz and the parietal portion Cz with the direction up, and the left anterior pinna from the origin Fpz in the direction of the left anterior pin point T3 and the right anterior pin point T4, respectively. Auxiliary region B having a horizontal axis width of approximately 4% between point T3 and right anterior pinna point T4, and 10% vertically between nasal root Nz and parietal portion Cz with the top direction Cz from the origin Fpz Axis width and left auricular anterior point T3 from the origin Fpz Auxiliary part C1 having a horizontal axis width of 10% between left anterior pinna point T3 and right anterior pinna point T4 in the direction of right anterior pin point T4, or nasal root portion Nz in the direction of parietal portion Cz from origin Fpz. From 7.5% to 20% of the vertical axis width between the parietal portions Cz and 7.5% between the left anterior pinna point T3 and the right anterior pin point T4 in the direction from the origin Fpz to the left anterior pinna point T3 A user terminal characterized in that the user terminal is any one of auxiliary portions C2 having a horizontal axis width of up to approximately 20%.

  (5) A prediction model that predicts a user's preference relationship from brain activity information of a specific part is stored, and is connected to be able to communicate with a server terminal that transmits additional information about the user's preference information using the prediction model A non-invasive brain activity measuring unit that measures brain activity information for determination of the specific part when specific information is presented, and the determination brain measured by the brain activity measuring unit A communication unit that transmits activity information to the server terminal, the communication unit receiving the additional information from the server terminal, and a presentation device that presents the received additional information to a user, , The vertical axis width from the lower 10% to the upper 30% between the nasal root Nz and the parietal portion Cz with the top of the parietal Cz direction from the origin Fpz in the international 10-20 method, and the left anterior pinna point from the origin Fpz 3 and the right anterior point T4 in the direction of the left auricular anterior point T3 and the right anterior point T4, an auxiliary part B having a width of about 4% between the anterior point T4 and the apex Cz direction from the origin Fpz. The vertical axis width of 10% below the nasal root portion Nz and the parietal portion Cz, and the left anterior pin point T3 and the right pin point in the direction from the origin Fpz to the left anterior pin point T3 and the right pin point T4, respectively. Auxiliary part C1 having a horizontal axis width of 10% between T4, or a vertical axis width from 7.5% to 20% above the nose root Nz / top Cz in the direction of the crown Cz from the origin Fpz, and the origin From Fpz in the direction of the left auricular point T3, the auxiliary part C2 having a horizontal axis width of 7.5% to approximately 20% between the left auricular point T3 and the right auricular point T4. A user terminal characterized by.

  (6) The user terminal according to (4) or (5), wherein the specific portion is 20% longitudinally between the nose root Nz and the parietal portion Cz in the direction of the parietal portion Cz from the origin Fpz in the international 10-20 method. Main axis consisting of an axial width and a horizontal axis width of approximately 4% between the left anterior pinna point T3 and the right anterior pinna point T4 in the direction from the origin Fpz to the left pinna point T3 and the right pinna point T4, respectively. A user terminal characterized by being a part A.

  (7) A prediction model construction system for predicting a user's preference relationship with respect to presented information, a non-invasive brain activity measurement unit that measures brain activity information of a specific part, and measurement by the brain activity measurement unit An analysis unit that builds a prediction model of the preference relationship from the brain activity information that has been obtained, and the specific site is the nose root Nz and the parietal region with the top Cz direction as the top from the origin Fpz in the International 10-20 method The vertical axis width from the lower 10% to the upper 30% between Cz, and between the left anterior pinna point T3 and the right pinna point T4 in the direction from the origin Fpz to the left anterior pin point T3 and the right pin point T4, respectively. Auxiliary region B having a horizontal axis width of approximately 4% of each, a vertical axis width of 10% below between the nose root Nz and the parietal portion Cz from the origin Fpz to the parietal portion Cz direction, and the left auricle from the origin Fpz In the direction of the front point T3 and right anterior pin point T4 Auxiliary part C1 having a horizontal axis width of 10% between the left anterior pinna point T3 and the right pinna point T4, or 7.5% between the nasal root portion Nz and the parietal portion Cz in the direction of the parietal portion Cz from the origin Fpz The vertical axis width from 20% to 20%, and the horizontal axis width from 7.5% to approximately 20% between the left anterior pinna point T3 and the right pinna point T4 in the direction from the origin Fpz to the left pinna point T3, A prediction model construction system, characterized in that the prediction model construction system is any one of the auxiliary parts C2.

  (8) A preference information presentation system that presents additional information about information preferred by the user using the prediction model constructed by the prediction model construction system according to (7), wherein the brain activity measurement unit Whether the user prefers the presented information from the judgment brain activity information and the prediction model measured by the brain activity measurement unit by measuring the brain activity information for judgment of the specific part when the information is presented A preference information presentation system comprising: a determination unit that determines whether or not, and a presentation device that presents additional information about the information on the condition that the determination unit determines that it is preferable.

  (9) The specific part is defined by the vertical width of 20% between the nose root Nz and the parietal portion Cz in the direction of the parietal portion Cz from the origin Fpz in the international 10-20 method, and the left anterior pinna point T3 and the right from the origin Fpz. Prediction according to (7), characterized in that it is a main part A having a horizontal axis width of approximately 4% between the left anterior pinna point T3 and the right anterior pinna point T4 in the direction of the anterior pinna point T4. A model construction system or a preference information presentation system according to (8).

  According to the user terminal described in (4) or the prediction model construction system described in (7), the same effect as the method described in (1) can be obtained.

  According to the user terminal described in (5) or the preference information presentation system described in (8), it is possible to achieve the same effect as the method described in (2).

  According to the user terminal described in (6) or the system described in (9), the same effect as the method described in (3) can be obtained.

  According to the present invention, it is possible to provide a method and an apparatus that perform prediction of a user's preference relationship unconsciously while suppressing introduction costs and maintenance costs, and that do not apply physical and mental stress to a subject.

It is a block diagram which shows the prediction model construction system of this invention. It is a figure which shows an example of the goods presentation screen shown in the case of prediction model construction. It is a figure which shows a NIRS mounting position. It is a figure which shows the international 10-20 law. It is a figure which shows the international 10-20 law. It is a figure which shows the change of the blood oxygenated hemoglobin density | concentration. It is a figure which shows the change of the blood oxygenated hemoglobin density | concentration. It is a figure which shows the change of the blood oxygenated hemoglobin density | concentration. It is a figure which shows the change of the blood oxygenated hemoglobin density | concentration. It is a figure which shows the relationship between blood oxygenated hemoglobin concentration and goods selection probability. It is a figure which shows the goods selection decision-making prediction by 3 layer perceptron. It is a figure which shows the goods selection decision-making prediction by 3 layer perceptron. It is a block diagram which shows a preference information presentation system. It is a figure which shows the flowchart for presenting preference information. It is a figure which shows an example of a brain activity measurement part. It is a figure which shows the main site | part regarding prediction of a preference relationship. It is a figure which shows the auxiliary | assistant part regarding prediction of a preference relationship. It is a figure which shows the auxiliary | assistant part regarding prediction of a preference relationship.

  Embodiments of the present invention will be described below with reference to the drawings.

[Overview of building a product selection decision-making prediction model]
In order to construct a prediction model for predicting a preference relationship between products from changes in cerebral blood oxyhemoglobin concentration at the time of product presentation, the following experiment was conducted. FIG. 1 is a block diagram showing a prediction model construction system 1.

  The prediction model construction system 1 includes a user terminal 1a including a brain activity information measurement unit 14 that measures brain activity information (for example, cerebral blood oxyhemoglobin concentration) of the user X, and a brain measured by the brain activity information measurement unit 14. A server terminal 1b that constructs a product selection decision-making prediction model based on activity information is provided. Note that the user terminal 1a and the server terminal 1b are connected so as to be able to communicate with each other.

  The user terminal 1a includes a presentation device 12, a selection unit 13, and a brain activity measurement unit 14. The server terminal 1 b includes a product information database 10, a presentation image determination unit 11, a brain information preprocessing unit 15, a subject information database 16, and an analysis unit 17. The user terminal 1a and the server terminal 1b are each provided with a communication unit capable of transmitting and receiving, but are not shown in order to clarify the data communication source and communication destination.

  The product information database 10 stores product data about products used for comparison. The presentation image determination unit 11 determines, from the product data stored in the product information database 10, product data in which two products are always of the same type, such as food, clothing, and electrical products, and presents the product data to the presentation device 12. Send to. Note that the product data transmitted to the presentation device 12 is also transmitted to the subject information database.

  The presentation device 12 presents the received two products to the user X. It should be noted that it is desirable to adjust so that the prices of the two products used for comparison are substantially the same. Here, FIG. 2 is a diagram showing a product presentation screen in the presentation device 12. In this experiment, the rest screen (a) was first presented, the product 1 screen (b) was presented, and then the product 2 screen (c) was presented for the same time as the product 1 screen (b) was presented. . The presentation time for the product 1 screen (b) and the product 2 screen (c) is about 9 seconds. Then, after the product 1 screen (b) and the product 2 screen (c) are presented, the product selection screen (d) is presented, and the selection of a more preferable product from the user X is accepted.

  When the product presentation screen as shown in FIG. 2 is presented, the user X uses the selection unit 13 to select a more preferable product. Note that there are three types of user X selections: “Product 1 is preferable”, “Product 2 is preferable”, and “Indiscriminate”. When the user X selects one of the three types, selection information indicating the selection result is transmitted to the subject information database 16.

  On the other hand, the brain activity measuring unit 14 measures the brain activity information of the user X when the product presentation screen is presented by the presentation device 12. Here, the concentration change of blood oxygenated hemoglobin is used for the construction of the prediction model. Moreover, in order to measure such brain activity information, NIRS (Near InfraRed Spectroscopy) which has a low spatial resolution and can only observe a low depth is used.

  NIRS is a non-invasive device that measures the amount of change in oxygenated hemoglobin and deoxygenated hemoglobin contained in blood using highly transmissive near-infrared light (wavelength 700 to 1000 nm). The measurement is performed by a pair of light transmitting points and light receiving points installed on the scalp at intervals of 3 cm, and light is absorbed while near infrared light irradiated from the light transmitting points is detected by the light receiving points. The amount of hemoglobin change in the blood is calculated backward from the amount of light, that is, the amount of attenuation of light reaching the light receiving point. NIRS is an apparatus suitable for experiments involving electronic devices because it is not affected by electronic noise in principle.

  Here, in order to pay attention to the part considered to reflect the activity of the reward system more, as shown in FIG. 3, in this experiment, the NIRS is attached to the frontal part, and the light transmitting point and the light receiving point are arranged in a grid pattern. By lining up, 32 channels were measured simultaneously. In FIG. 3, a black character portion in the white circle indicates a light transmission point, a white character portion in the black circle indicates a light receiving point, and a black character portion in a white square indicates a channel. The NIRS was mounted using the International 10-20 method (see FIGS. 4 and 5), and the sampling period was 110 ms.

  Referring to FIG. 1 again, the brain information preprocessing unit 15 performs preprocessing such as smoothing to remove periodic fluctuations peculiar to the living body from the brain activity information measured by the brain activity measurement unit 14. . The brain activity information smoothed by the brain information preprocessing unit 15 is transmitted to the subject information database 16.

  The subject information database 16 includes product data determined to be presented by the presentation image determination unit 11, selection information selected by the user X when the product is presented, and brain activity information when the product is presented (smoothed). , The product data, selection information, and brain activity information are transmitted to the analysis unit 17. The analysis unit 17 constructs a product selection decision-making prediction model using the measured brain activity information as a prediction factor, and transmits it to the subject information database 16.

  Thus, the server terminal 1b constructs a product selection decision-making prediction model based on the brain activity information measured by the user terminal 1a. Therefore, it can be said that the user terminal 1a is communicably connected to the server terminal 1b that constructs the product decision prediction model based on the brain activity information.

[Experimental result]
6 to 9 are plots of changes in the blood oxygenated hemoglobin concentration at each site when product 1 is selected (a) and when product 2 is selected (b) for each subject. In each figure, the vertical axis represents the difference in blood oxygenated hemoglobin concentration, and the horizontal axis represents the time since the product image was presented. The difference in cerebral blood oxyhemoglobin concentration is calculated by “(cerebral blood oxyhemoglobin concentration at the time of product 1 image presentation) − (cerebral blood oxyhemoglobin concentration at the time of product 2 image presentation)” for each subject. The average is plotted.

  FIG. 6 shows changes in cerebral blood oxygenated hemoglobin concentration in channel 10 (10 ch), and FIG. 7 shows changes in cerebral blood oxygenated hemoglobin concentration in channel 23 (23 ch). FIG. 8 shows changes in cerebral blood oxygenated hemoglobin concentration in channel 16 (16ch), and FIG. 9 shows changes in cerebral blood oxygenated hemoglobin concentration in channel 17 (17ch). That is, FIGS. 6 to 9 show changes in cerebral blood oxyhemoglobin concentration of 10 ch, 23 ch, 16 ch, and 17 ch.

  According to FIG. 6 to FIG. 9, clear symmetry is seen in changes in cerebral blood oxyhemoglobin concentration at 10 ch and 23 ch. Specifically, according to (a) of FIG. 6 and (a) of FIG. 7, when the product 1 is selected, the difference in cerebral blood oxyhemoglobin concentration increases in the positive direction as time elapses. . On the other hand, according to (b) of FIG. 6 and (b) of FIG. 7, when the product 2 is selected, the difference in cerebral blood oxyhemoglobin concentration increases in the negative direction as time elapses.

  In 16ch and 17ch, there is a certain symmetry in the change in cerebral blood oxyhemoglobin concentration when product 1 is selected and when product 2 is selected (see FIGS. 8 and 9). .

  Here, the difference in cerebral blood oxyhemoglobin concentration is calculated by “(cerebral blood oxyhemoglobin concentration when the product 1 image is presented) − (cerebral blood oxyhemoglobin concentration when the product 2 image is presented)”. . Therefore, when the difference increases in the positive direction, it means that the oxygenated hemoglobin concentration in the cerebral blood when the product 1 image is presented is high, and when the difference increases in the negative direction, the brain when the product 2 image is presented. It means that blood oxygenated hemoglobin concentration is high. As a result, it is found that when the product selected by the user (the product preferred by the user) is presented, the cerebral blood oxyhemoglobin concentration in 10ch, 23ch, 16ch, and 17ch increases. That is, when product 1 is selected (preferred), activation when viewing the image of product 1 is larger than when viewing the image of product 2, and conversely, when product 2 is selected (preferred) It can be seen that the activation when viewing the product 2 image is greater than when viewing the product 1 image.

  FIGS. 6 to 9 show changes in cerebral blood oxyhemoglobin concentrations at 10 ch, 23 ch, 16 ch, and 17 ch. However, changes in cerebral blood oxyhemoglobin concentrations at 29 ch and 30 ch also have a certain symmetry. It was.

  FIG. 10 shows the relationship between the change in cerebral blood oxyhemoglobin concentration in 10ch and the product selection probability using a quartile analysis.

  The horizontal axis represents the difference in the change in cerebral blood oxyhemoglobin concentration at 10 ch, “(change in oxyhemoglobin concentration when the product 1 image is presented) − (change in cerebral blood oxyhemoglobin concentration when the product 2 image is presented)”. However, since there is an individual difference in the distribution (average, variance) of changes in cerebral blood oxyhemoglobin concentration, the average is normalized to 0 and the variance is normalized to 1 for each subject to remove the individual difference. The vertical axis represents the probability of selecting product 1.

  In FIG. 10, regarding the difference in normalized oxyhemoglobin concentration change, the entire interval is divided into eight, and the ratio between the case where the product 1 is selected and the case where the product 1 is not selected among the samples included in each divided interval is a probability. It is said. In this figure, the stronger the activation of 10ch is observed when the product 1 is presented (compared to the presentation of the product 2) on the right side, while the 10ch of the 10ch channel is presented when the product 2 is presented (compared with the presentation of the product 1). A strong activation will be seen.

[Building a prediction model]
Based on these observation results, a product selection decision-making prediction model was constructed using the observed cerebral blood oxyhemoglobin concentration as a prediction factor in order to statistically support the observation results. The prediction model uses a three-layer perceptron, and parameters and hyperparameters are Bayesian estimated by the Markov chain Monte Carlo method (MCMC method). Hyperparameter marginal likelihood was used to select the intermediate layer and prediction factors. As predictive variables, in addition to factors that capture individual product purchase trends (using purchase probabilities immediately before decision making, hereinafter referred to as “autocorrelation terms”), each part corresponding to 10ch, 23ch, 16ch, and 17ch Changes in cerebral blood oxygenated hemoglobin concentration (FIGS. 6 to 9) were used. The predicted variable is the subject's product selection. The case where the product 1 is selected is “+1”, the product 2 is “−1”, and the indiscriminateness is “0”. However, both the predictor variable and the predictor variable are normalized for each subject's data, and differences in distribution among individuals are removed.

  As a result of model selection based on the hyperparameter marginal likelihood, the number of intermediate layers is 3, and the predictive factor is the autocorrelation term and the change in cerebral blood oxyhemoglobin concentration at 10ch and 23ch as a significant factor. Adopted. The hyperparameter marginal likelihood at this time is −236.6, and an improvement of about 24% is recognized in the hyperparameter marginal likelihood as compared with the prediction model having only the autocorrelation term. By adding changes in cerebral blood oxyhemoglobin concentration to the predictor variable, it can be said that the explanatory power of product selection behavior has increased significantly.

  FIG. 11 shows a prediction model in product selection. X-axis shows changes in cerebral blood oxyhemoglobin concentration at 10 ch when product 1 is viewed, Y-axis shows changes in cerebral blood oxyhemoglobin concentration at 10 ch when product 2 is viewed, Z-axis is selected (product 1 selection is “ +1 ”, product 2 selection is“ −1 ”, and indiscriminate is“ 0 ”, normalized by each subject).

  Referring to FIG. 11, in the prediction model, the positive change in the cerebral blood oxygenated hemoglobin concentration when the product 1 is presented is larger, and the more the negative change in the cerebral blood oxygenated hemoglobin concentration when the product 2 is presented, the greater the product 1 is. Predicted to be selected. Conversely, the negative change in cerebral blood oxyhemoglobin concentration when product 1 is presented is large, and the greater the positive change in cerebral blood oxyhemoglobin concentration when product 2 is presented, the more predicts that product 2 will be selected. ing.

  Similarly, FIG. 12 shows a prediction model of a three-layer perceptron, but it is a case where not the cerebral blood oxyhemoglobin concentration at the time of each product presentation but a difference between them is used as a prediction variable. As a result of model selection, a three-layer perceptron having four intermediate layers was adopted, and the hyperparameter marginal likelihood was −248.0. The horizontal axis of the figure is the difference in cerebral blood oxyhemoglobin concentration change at 10 ch when each product is presented (product 1-product 2), and the vertical axis is selected (product 1 selection is "+1", product 2 selection is "-1") Indiscriminateness is “0”). However, the data of each subject is normalized in advance. Similarly to FIG. 11, from FIG. 12, product 1 is selected as the change in cerebral blood oxyhemoglobin concentration when product 1 is presented is larger than the change in cerebral blood oxyhemoglobin concentration when product 2 is presented. It can be predicted that.

  Here, the knowledge about the user's decision making in the past has been obtained by measuring a relatively deep part of the brain using MRI having a high spatial resolution (for example, a reward system or an emotion system). On the other hand, when using a simple measuring device with low spatial resolution and low depth, there are many uncertainties as to how much knowledge can be obtained about user decision making. About this point, it became clear by this experiment that even if it is a case where NIRS which is a convenient measuring device with low spatial resolution and low depth is used, it is possible to predict a user's decision making.

  As a result, it is possible to predict user preferences at relatively low introduction costs and maintenance costs without using a large-scale facility such as MRI. Moreover, since the site | part to measure is specified, a user's preference can be estimated, relieving a user's physical and mental stress, without imposing a body movement restriction | limiting on a user.

  In particular, since the identified part is a suitable part for the prediction of the user's preference, it can be expected that the accuracy of the prediction is significantly increased. In addition, since the preference of the user is predicted using brain information, unconscious or emotional preference can also be measured quantitatively.

[Overview of preference information presentation system]
The preference information presentation system 2 using the prediction model constructed as described above will be described with reference to FIG.

  The preference information presentation system 2 includes a user terminal 2a including a brain activity information measurement unit 14 that measures brain activity information (for example, cerebral blood oxyhemoglobin concentration) of the user X, and a brain measured by the brain activity information measurement unit 14. The server terminal 2b which determines the user's X preference based on activity information is provided. The user terminal 2a and the server terminal 2b are connected so as to be able to communicate with each other.

  The user terminal 2 a includes at least a presentation device 12 and a brain activity measurement unit 14. The server terminal 2b includes at least a product information database 10, a presentation image determination unit 11, a brain information preprocessing unit 15, a subject information database 16, a determination unit 20, and an auxiliary information creation unit 21. Here, the prediction model is stored in the subject information database 16. In addition, although the user terminal 2a and the server terminal 2b are each provided with the communication part which can be transmitted / received, in order to clarify the communication origin and communication destination of data, illustration is abbreviate | omitted.

  The product information database 10 stores product data presented to the presentation device 12 and auxiliary data related to the product data. Here, the auxiliary data refers to data related to product data such as more detailed data about the product or data about a product similar to the product.

  For example, the presentation image determination unit 11 determines product data to be presented to the presentation device 12 in accordance with a selection from the user X. The presentation device 12 presents the determined product data to the user X.

  The brain activity measuring unit 14 measures the brain activity information of the user X when the product presentation screen is presented by the presentation device 12. The measured brain activity information is transmitted to the brain information preprocessing unit 15. The brain information preprocessing unit 15 performs preprocessing such as smoothing on the measured value. Then, the measurement value smoothed by the brain information preprocessing unit 15 is transmitted to the determination unit 20.

  The determination unit 20 takes out the prediction model stored in the subject information database 16 and determines whether or not the user X likes the product presented by the prediction model and the measurement value.

  If the auxiliary information creating unit 21 determines that it is preferred by the determining unit 20, the auxiliary information creating unit 21 extracts auxiliary data related to the presented product from the product information database 10 and transmits it to the presentation device 12. The presentation device 12 presents the received auxiliary data to the user X.

  Thus, the server terminal 2b determines the preference of the user X based on the brain activity information measured by the user terminal 2a. Therefore, it can be said that the user terminal 2a is communicably connected to the server terminal 2b that determines the preference of the user X based on the brain activity information.

  In addition, in the preference information presentation system 2, although the information which the user X likes is shown, you may provide the structure which builds a prediction model with this. That is, as shown in FIG. 13, the prediction model may be constructed as described above by providing the selection unit 13 in the user terminal 2a and the analysis unit 17 in the server terminal 2b. As a result, the prediction model can be updated as appropriate.

[Flowchart for Presenting Preference Information]
With reference to FIG. 14, the flowchart for presenting preference information by the preference information presentation system 2 will be described.

  First, a stimulus such as product data is presented to the user via the presentation device 12 (S101). Subsequently, a change in cerebral blood oxyhemoglobin concentration at a specific site is measured via the brain activity measuring unit 14 (S102).

  Subsequently, the measured value is smoothed through the brain information preprocessing unit 15 (S103). Subsequently, the smoothed measurement value is input to the prediction model via the determination unit 20 (S104), and it is determined whether or not this measurement value exceeds the threshold value of the prediction model (S105).

  If this determination does not exceed the threshold value, it is determined that the user does not like the product, and the process ends. On the other hand, if it determines with exceeding the threshold value, the determination part 20 will determine with liking (S106). Subsequently, auxiliary data related to the product preferred by the user is created from the product information database 10 via the auxiliary information creating unit 21, and this auxiliary data is presented to the presentation device 12 (S107). Thereafter, the process ends.

  In this way, the preference information presentation system 2 can automatically present additional information about information preferred by the user. As a result, for example, when a user is looking at a shopping site via the Internet, it is possible to predict the preference from the user's brain information and to give additional information on a product that the user likes, which is highly practical. can do.

  In the present embodiment, the stimulus is presented by presenting an image to the user (S101), but the present invention is not limited to this. That is, an auditory stimulus may be given instead of a visual stimulus. For example, it is good also as predicting a music preference from the brain information at the time of a user viewing specific music on a music site via the Internet, and providing additional information of music preferred by the user.

[Brain Activity Measurement Unit]
According to this embodiment, since the site | part which measures brain information is specified, size reduction of the brain activity measurement part 14 can be expected. Accordingly, an example of a suitable brain activity measurement unit 14 of the present embodiment will be described with reference to FIG.

  The brain activity measuring unit 14 includes a band unit 200, a measuring unit 201, and a brain information transmitting unit (not shown). The measuring unit 201 includes a light transmitting unit 201a and a light receiving unit 201b, and measures user brain information. The light transmitting unit 201a and the light receiving unit 201b measure brain information in a site (for example, 10ch) suitable for predicting a preference relationship. The band unit 200 is used to fix the measurement unit 201 at a suitable position for measuring brain information.

  As described above, according to the present embodiment, the apparatus can be reduced in size, the body movement restriction is not imposed on the user, and the user's preference can be determined at a relatively low introduction cost and maintenance cost. . In particular, since the measurement unit 201 specifies the part to be measured, it is possible to accurately determine the user's preference.

[Identification of measurement site]
Below, the suitable site | part for predicting (determining) a preference relationship is demonstrated with reference to FIGS.

  In addition, in FIGS. 16-18, a measurement site | part is identified using the description (refer FIG.4 and FIG.5) in the international 10-20 method. 16 to 18, the length of the vertical axis from the nasal root Nz to the top of the head Cz is 20 cm, and the length of the horizontal axis from the left anterior pinna point T3 to the right anterior pin point T4 is 30 cm. Assuming that the measurement site is specified. Further, the direction from the origin Fpz toward the crown Cz is “up”, the direction from the origin Fpz to the nasal root Nz is “down”, the direction from the origin Fpz to the left anterior pinna T3 is “left”, and the origin Fpz The direction from the right to the right anterior point T4 is expressed as “right”.

  With reference to FIG. 16, the most suitable main part A regarding the prediction (determination) of a preference relationship is demonstrated.

  The main part A has a width of 4 cm with the vertical axis 4 cm above the origin Fpz. This is a range from 20% to 40% from the nasal root Nz (0%) to the top of the head Cz (100%), that is, a range of 20% above the origin Fpz. The horizontal axis is 2.5 cm wide from the origin Fpz to the left and right by 1.25 cm. This is a range from about 46% to about 54% from the left anterior pinna point T3 (0%) to the right pinna point T4 (100%), that is, the left pinna point from the origin Fpz to the left and right respectively. The range is approximately 4% between T3 and the right anterior pinna point T4.

  As described above, the most preferable main part A has a range of 20% above the origin Fpz and a range of approximately 4% between the left anterior pinna point T3 and the right anterior pinna point T4 to the left and right from the origin Fpz. It is a site. The main part A is a part including the 10ch, 23ch, 16ch, and 17ch described above.

  As described above, the most suitable site for the prediction (determination) of the preference relationship has been described. However, there are auxiliary sites other than these sites that can achieve a certain accuracy for the prediction (determination) of the preference relationship. Hereinafter, with reference to FIG. 17 and FIG. 18, an explanation will be given on a suitable auxiliary part for prediction (determination) of the preference relationship.

  FIG. 17 shows an auxiliary part B capable of realizing a certain accuracy regarding the prediction (determination) of the preference relationship.

  The auxiliary part B has a width of 8 cm from the lower 2 cm to the upper 6 cm of the origin Fpz on the vertical axis. This is a range from 10% to 50% from the nasal root Nz (0%) to the top of the head Cz (100%), that is, a range from 10% below the origin Fpz to 30% above. As for the horizontal axis, the same horizontal axis as the main part A is 2.5 cm wide by 1.25 cm from the origin Fpz to the left and right. This is a range from about 46% to about 54% from the left anterior pinna point T3 (0%) to the right pinna point T4 (100%), that is, the left pinna point from the origin Fpz to the left and right respectively. The range is approximately 4% between T3 and the right anterior pinna point T4.

  As described above, the auxiliary part B has a range from the lower 10% to the upper 30% of the origin Fpz and approximately 4% between the left anterior pinnae point T3 and the right pinnae anterior point T4 respectively to the left and right from the origin Fpz. A region consisting of a range. Even in such an auxiliary part B, a certain degree of accuracy can be realized in the prediction of the preference relationship and the determination of the preference relationship.

  FIG. 18 shows an auxiliary part B, an auxiliary part C1, and an auxiliary part C2 that can achieve a certain degree of accuracy regarding the prediction (determination) of the preference relationship.

  The auxiliary part C1 has a width of 2 cm with a vertical axis of 2 cm below the origin Fpz. This is a range from 10% to 20% from the nasal root Nz (0%) to the crown Cz (100%), that is, a range of 10% below the origin Fpz. The horizontal axis is a width of 6 cm, 3 cm left and right from the origin Fpz. This is a range from 40% to 60% from the left anterior pinna point T3 (0%) to the right anterior pin point T4 (100%), that is, the left anterior pin point T3. The range is 10% between the right anterior pinna point T4.

  The auxiliary part C2 has a width of 2.5 cm from 1.5 cm to 4 cm above the origin Fpz on the vertical axis. This is a range from 27.5% to 40% from the nasal root Nz (0%) to the crown Cz (100%), that is, a range from 7.5% to 20% above the origin Fpz. The horizontal axis is a width of 3.5 cm from 2.25 cm to 5.75 cm on the left of the origin Fpz. This is a range from about 30% to about 42.5% from the left anterior pinna point T3 (0%) to the right pinna point T4 (100%), that is, from the origin Fpz to the left pinna point T3. The range is from 7.5% to approximately 20% between the left anterior pinna point T3 and the right pinna point T4 in the direction.

  As described above, the auxiliary part C1 is a part having a range of 10% below the origin Fpz and a 10% range between the left anterior pinna point T3 and the right pinna point T4 to the left and right from the origin Fpz. is there. Further, the auxiliary part C2 has a vertical axis in the range from 7.5% to 20% above the origin Fpz and a horizontal axis from the origin Fpz in the direction of the left auricular point T3 in front of the left auricular point T3 / in front of the right auricle. This is a region consisting of a range from 7.5% to about 20% between the points T4. Even in such auxiliary parts C1 and C2, a certain degree of accuracy can be realized in the prediction of the preference relationship and the determination of the preference relationship. The auxiliary part C1 is a part including the above-described 29ch and 30ch, and the auxiliary part C2 is a part including the above-described 11ch and 12ch.

[effect]
As described above, according to the present embodiment, since the user's preference is predicted (determined) using the brain activity information, unconscious or emotional preference can also be quantitatively measured.

  In addition, since the region to be measured is specified in the prediction (determination) using the brain activity information, the measurement device can be reduced in size (see FIG. 15), and the body movement is restricted on the user. The user's preference can be predicted (determined) at a relatively low introduction cost and maintenance cost. As a result, it can be applied to, for example, an Internet shopping site, and can be made practical.

  Moreover, the specified site | part is a site | part suitable for prediction (determination) of a preference relationship, and a preference relationship can be estimated (determination) with high precision.

[Others]
In the present embodiment, NIRS is used for measuring brain activity information, but the present invention is not limited to this. For example, a general apparatus such as EEG (Electroencephalogram) that is not invasive and does not limit body movement or experimental environment may be used. In other words, the present invention is characterized by measuring brain activity information of a specific part in predicting (determining) the selection relationship. If the apparatus is not invasive and has little body movement limitation, the measuring apparatus is appropriately used. It can be changed.

DESCRIPTION OF SYMBOLS 1 Prediction model construction system 2 Preference information presentation system 10 Merchandise information database 11 Presentation image determination part 12 Presentation apparatus 13 Selection part 14 Brain activity measurement part 15 Brain information pre-processing part 16 Test subject information database 17 Analysis part 20 Determination part 21 Auxiliary information creation Part

Claims (9)

A method of constructing a prediction model of user preference relationships for presented information comprising:
A step of measuring brain activity information of a specific part by a non-invasive brain activity measurement unit;
Constructing a prediction model of the preference relationship from the measured brain activity information,
The specific site,
In the international 10-20 method, the vertical axis width from the lower 10% to the upper 30% between the nasal root Nz and the parietal Cz from the origin Fpz to the parietal Cz direction, the left anterior pinna T3 and the right from the origin Fpz Auxiliary region B having a horizontal axis width of about 4% between the left anterior pinna point T3 and the right anterior pinna point T4 in the direction of the anterior pinna point T4.
The vertical axis width of 10% below the nose root portion Nz and the parietal portion Cz from the origin Fpz to the parietal portion Cz direction, and the left auricle from the origin Fpz in the direction of the left auricular anterior point T3 and the right auricular anterior point T4, respectively. Auxiliary region C1 consisting of a 10% horizontal axis width between the anterior point T3 and the right auricular anterior point T4,
Alternatively, the vertical axis width from the origin Fpz to the parietal portion Cz in the direction of the nasal root Nz and the parietal portion Cz is 7.5% to 20%, and the left anterior pinna point T3. An auxiliary part C2 having a horizontal axis width of 7.5% to approximately 20% between the right anterior pinna point T4,
Any one of the methods described above. A method for presenting additional information about information preferred by a user using a prediction model constructed by the method of claim 1, comprising:
Presenting specific information to the user;
Measuring the brain activity information for determining the specific part when the specific information is presented by the brain activity measuring unit;
Determining whether the user likes the presented information from the determination brain activity information and the prediction model;
Presenting additional information about the information on condition that it is determined to be preferred. The method according to claim 1 or 2, wherein
The specific site,
In the international 10-20 method, the vertical axis width of 20% between the nasal root Nz and the parietal portion Cz from the origin Fpz to the parietal portion Cz, and from the origin Fpz to the left anterior pinna point T3 and the right anterior pinna point T4, respectively. A method characterized in that a main part A having a horizontal axis width of about 4% between the left anterior pinna point T3 and the right pinna point T4 is used. A user terminal communicably connected to a server terminal capable of constructing a prediction model for predicting a user's preference relationship from the user's brain activity information,
A non-invasive brain activity measurement unit that measures brain activity information of a specific site;
A communication unit that transmits the brain activity information measured by the brain activity measurement unit to the server terminal,
The specific site,
In the international 10-20 method, the vertical axis width from the lower 10% to the upper 30% between the nasal root Nz and the parietal Cz from the origin Fpz to the parietal Cz direction, the left anterior pinna T3 and the right from the origin Fpz Auxiliary region B having a horizontal axis width of about 4% between the left anterior pinna point T3 and the right anterior pinna point T4 in the direction of the anterior pinna point T4.
The vertical axis width of 10% below the nose root portion Nz and the parietal portion Cz from the origin Fpz to the parietal portion Cz direction, and the left auricle from the origin Fpz in the direction of the left auricular anterior point T3 and the right auricular anterior point T4, respectively. Auxiliary region C1 consisting of a 10% horizontal axis width between the anterior point T3 and the right auricular anterior point T4,
Or, the vertical axis width from 7.5% to 20% between the nose root Nz and the parietal portion Cz in the direction from the origin Fpz to the parietal portion Cz, and the left anterior pinna point T3 · in the direction from the origin Fpz to the left anterior pinpoint T3. An auxiliary part C2 having a horizontal axis width of 7.5% to approximately 20% between the right anterior pinna point T4,
Any one of the above-mentioned user terminals. A user terminal that is connected to a server terminal that is communicable with a server terminal that stores a prediction model in which a user's preference relationship is predicted from brain activity information of a specific part and transmits additional information about the user's preference information using the prediction model. There,
A non-invasive brain activity measurement unit that measures the brain activity information for determination of the specific part when the specific information is presented;
A communication unit that transmits the determination brain activity information measured by the brain activity measurement unit to the server terminal, the communication unit receiving the additional information from the server terminal;
A presentation device for presenting the received additional information to a user,
The specific site,
In the international 10-20 method, the vertical axis width from the lower 10% to the upper 30% between the nasal root Nz and the parietal Cz from the origin Fpz to the parietal Cz direction, the left anterior pinna T3 and the right from the origin Fpz Auxiliary region B having a horizontal axis width of about 4% between the left anterior pinna point T3 and the right anterior pinna point T4 in the direction of the anterior pinna point T4.
The vertical axis width of the bottom 10% between the nose root Nz and the parietal portion Cz from the origin Fpz to the parietal portion Cz direction, and the left auricle from the origin Fpz in the direction of the left auricular anterior point T3 and the right auricular anterior point T4. Auxiliary region C1 consisting of a 10% horizontal axis width between the anterior point T3 and the right auricular anterior point T4,
Alternatively, the vertical axis width from the origin Fpz to the parietal portion Cz in the direction of the nasal root Nz and the parietal portion Cz is 7.5% to 20%, and the left anterior pinna point T3. An auxiliary part C2 having a horizontal axis width of 7.5% to approximately 20% between the right anterior pinna point T4,
Any one of the above-mentioned user terminals. The user terminal according to claim 4 or 5, wherein
The specific site,
In the international 10-20 method, the vertical axis width of 20% between the nasal root Nz and the parietal portion Cz from the origin Fpz to the parietal portion Cz, and from the origin Fpz to the left anterior pinna point T3 and the right anterior pinna point T4, respectively. A user terminal characterized by a main part A having a horizontal axis width of about 4% between the left anterior pinna point T3 and the right pinna point T4. A prediction model construction system for predicting a user's preference relationship for presented information,
A non-invasive brain activity measurement unit that measures brain activity information of a specific site;
An analysis unit that builds a prediction model of the preference relationship from the brain activity information measured by the brain activity measurement unit,
The specific site,
In the international 10-20 method, the vertical axis width from the lower 10% to the upper 30% between the nasal root Nz and the parietal Cz from the origin Fpz to the parietal Cz direction, the left anterior pinna T3 and the right from the origin Fpz Auxiliary region B having a horizontal axis width of about 4% between the left anterior pinna point T3 and the right anterior pinna point T4 in the direction of the anterior pinna point T4.
The vertical axis width of 10% below the nose root portion Nz and the parietal portion Cz from the origin Fpz to the parietal portion Cz direction, and the left auricle from the origin Fpz in the direction of the left auricular anterior point T3 and the right auricular anterior point T4, respectively. Auxiliary region C1 consisting of a 10% horizontal axis width between the anterior point T3 and the right auricular anterior point T4,
Alternatively, the vertical axis width from the origin Fpz to the parietal portion Cz in the direction of the nasal root Nz and the parietal portion Cz is 7.5% to 20%, and the left anterior pinna point T3. An auxiliary part C2 having a horizontal axis width of 7.5% to approximately 20% between the right anterior pinna point T4,
A prediction model construction system characterized by being one of the following. A preference information presentation system for presenting additional information about information preferred by a user using a prediction model constructed by the prediction model construction system according to claim 7,
The brain activity measurement unit measures the brain activity information for determination of the specific part when specific information is presented,
A determination unit that determines whether or not a user likes the presented information from the determination brain activity information measured by the brain activity measurement unit and the prediction model;
And a presentation device that presents additional information about the information on condition that the determination unit determines that the preference information is preferable. The specific site,
In the international 10-20 method, the vertical axis width of 20% between the nasal root Nz and the parietal portion Cz from the origin Fpz to the parietal portion Cz, and from the origin Fpz to the left anterior pinna point T3 and the right anterior pinna point T4, respectively. 8. The prediction model construction system according to claim 7, or a main part A having a horizontal axis width of about 4% between the left anterior pinna point T3 and the right pinna point T4. Preference information presentation system described.

Images