JP5382943B2 - Preference determination method and preference determination program - Google Patents

Description

  The present invention relates to a method for determining human preference and a program for determining human preference.

  The preferences include those related to products and services, those related to the content of stories, and those related to occupation. Determining the presence or intensity of these preferences is important when recruiting companies, lecturers, teachers, or job seekers who are interested in selling products and services to consumers. It is important for companies and government agencies that identify job preferences of job seekers.

  Taste is a kind of human mental activity. Human mental activity is classified into three areas: a cognitive area (head), an emotional area (heart), and a psychomotor area (body) (for example, see Non-Patent Document 1). The cognitive domain is "knowledge", "understanding", "application", "analysis", "integration", and "evaluation", and the emotional domain is "acceptance", "reaction", "value", "organization", and "individualization" In other words, the psychomotor area is organized as an area relating to the acquisition of motor movement (for example, see Non-Patent Document 2). Preference or synonymous interest is a mental activity that belongs to the emotional domain.

  Conventionally, there are three types of methods for determining human preference. The first method is a questionnaire examination, the second method is an interview examination, and the third method is a pupil examination.

  Questionnaire inspection and interview inspection are methods for examining subjective feelings and are called psychological measures. Regarding the measurement of emotions using a psychological scale, it has been concluded that "the influence of intentional or unconscious lies and concealment cannot be avoided" (for example, see Non-Patent Document 3).

  For example, there is a Japanese version of the VPI occupation interest examination in the questionnaire examination for the purpose of determining the preference for occupation (for example, see Non-Patent Document 4). According to Non-Patent Document 5, the model proposed by Holland based on the Japanese version of the VPI occupational interest examination has had a great influence as a theory supporting the occupational interest examination for many years.

  The third method, pupil examination, is described in Patent Document 1 and Non-Patent Document 6. Patent Document 1 describes a system that acquires sensitivity information from a pupil and presents an image according to the sensitivity information. Non-Patent Document 6 describes determination of interest in a television program by pupil area.

  In Non-Patent Document 7, the mental load in the case where a subject is memorized as a numerical string is determined based on the pupil area. Non-Patent Document 7 clarifies that the maximum value of the pupil area increases as the number of digits to be stored increases in the memory task of memorizing, holding, and reproducing a 3 to 7 digit number string.

Japanese Patent Application Laid-Open No. 11-282721

Junichi Hamada, “Education Evaluation”, Yuyakukaku, published in 1983 Katsuaki Suzuki, Aya Watanabe, “Detailed Taxonomy Classification and Test Item Examples”, [online], December 18, 2009, Kumamoto University Graduate School, [Search December 18, 2009], Internet <URL: http: // www. gsis. kumamoto-u. ac. jp / opencourses / pf / 2Block / 04 / 04-taxonomic-hyou. html> Tsujiyo, Naoto Suzuki, Yasuhisa Tsuji, “Invitation to Emotional Psychology -Approach to Emotions and Emotions”, Science, 2001, pp. 10 Labor Policy Research and Training Organization, “Occupational Information and Employment Support Tool”, [online], December 18, 2009, Labor Policy Research and Training Organization, [December 18, 2009 search] ] <Internet: <URL: http: // www. jil. go. jp / institute / seika / tools / VPI. htm> Naoto Watanabe, “Development of a new vocational interest model: Beyond the RIASEC model”, Proceedings of the Annual Meeting of the Japan Society for Management Behavioral Science (5), published on November 5, 2002, pp. 103-111 Masaki Murai, Minoru Nakayama, Yasutaka Shimizu, “Relationship between Pupil Area and Interest and Impression of Image Contents when Watching TV Programs”, The Journal of the Institute of Image Information and Television Engineers, vol. 52, no. 11, published in 1998, pp. 1748-1753 J. et al. Beatty, “Task-evoked public response, processing load, and the structure of processing resources,” Psychological Bulletin, vol. 91, no. 2, 1982, pp. 276-292

  Since the pupil area changes depending on the reaction of the nervous system, the influence of lies and concealment is negligible, as in questionnaires and interviews. Thus, from the viewpoint of preference determination accuracy, the use of examination based on the pupil area is expected in the future. However, the conventional pupil examination method has two problems.

  The first problem is that the process of analyzing the pupil area is unclear. For example, in Patent Document 1, it is estimated that a sensitivity state that a pupil has a preference if it shows a reaction, “when image information (omitted) that the pupil opens is directly input as a sensitivity information source” As a result, the sensibility information determination unit 3 obtains the sensibility information of interest ”, and the specific analysis process is not clear. Since it is unclear how to analyze the data obtained from the living body, which is the pupil area, it cannot be said to be feasible.

  The second problem is that it is extremely difficult to control the brightness, which causes a problem in the accurate determination of preferences. It is already known that the pupil area is more strongly affected by luminance than preference. The pupil has a physiological function to adjust the amount of light that is projected onto the retina. When the luminance is high, the pupil is reduced to reduce the amount of light that is projected onto the retina, and conversely, when the luminance is low, the pupil is enlarged. To increase the amount of light projected on the retina. According to Non-Patent Document 6, an interest measurement based on a pupil area uses an image of a person or a landscape or a TV program video, and a specific image or video of interest is selected from a plurality of images or videos according to the pupil area. A method of determination is described. Here, if the luminances of the plurality of images and moving images are different from each other, the pupil area is affected by the luminance rather than the interest, and the interest that is the purpose of the original analysis cannot be determined. In Non-Patent Document 6, it is defined that the pupil area P is defined by two additions of the luminance B (x) and the mental activity S (y), and the influence of the luminance B (x) on P is calculated after the end of the experiment, We are trying to calculate the effect on pupil area due to S (y) alone. In Non-Patent Document 6, various television program videos (for example, two videos of a news program and a sports program) are presented in front of the subject's eyes, but the video is composed of a large number of pixels having various luminances. It is fundamentally impossible to unify the brightness between different videos. In Non-Patent Document 6, the average value of the luminance of the entire screen is calculated from the luminance values of 432 lattice points on the screen for each moving image, and is substituted for the luminance of the moving image. While trying to remove the influence and trying to extract the influence on the pupil area only by the mental activity of interest, on the screen where the luminance dispersion of each grid point is large, the method according to Non-Patent Document 6 uses the true The brightness cannot be captured. As a result, there is a possibility that the judgment accuracy of the mental activity of interest is inferior. Since not only a moving image but also a person or landscape image is composed of a large number of pixels having different luminances, the same possibility appears.

  As described above, the present invention intends to solve the problems of the questionnaire inspection and the interview inspection and the pupil inspection in the conventional preference determination, and an object thereof is to provide a preference determination method with little room for deception. It is what.

  In order to achieve the above object, the pupil area of the subject is measured when viewing each of the target screen and the non-target screen on which the character string is displayed. And the average value of each pupil area was compared and it was decided to determine the presence or absence of a test subject's preference.

The preference determination system according to claim 1 is at least one or more character strings included in a character string group that is a collection of character strings related to preference, and calculates the familiarity of the character strings in advance to a certain range. When viewing the target screen, a presentation means for presenting to the subject a target screen in which the corresponding character string is written, and a non-target screen that is a blank that has a character string that is not included in the character string group or that is not written. Measuring means for measuring the target screen pupil area that is the pupil area of the subject and the non-target screen pupil area that is the pupil area of the subject when viewing the non-target screen, and the target screen pupil measured by the measuring means A test means for statistically testing a difference between the area and the non-target screen pupil area, and the presentation means is the intimacy performed for each subject in advance. Based on the findings regarding, closeness is selected more strings threshold, characterized in that a character string corresponding to the character string that is the selected in the predetermined range.

  According to this method, by controlling the word familiarity to a certain level or more, the influence of cognitive mental activity on the pupil area can be suppressed, and the preference determination can be realized with high accuracy.

The preference determination system according to claim 2 is at least one or more character strings included in a character string group that is a collection of character strings related to preference, and calculates the familiarity of the character strings in advance to a certain range. A presenting means for presenting a target screen representing a corresponding character string and a non-target screen representing a character string not included or not represented in the character string group to a subject in a range of 0.5 seconds to less than 5 seconds; Measuring means for measuring a target screen pupil area that is a pupil area of a subject when viewing the target screen and a non-target screen pupil area that is a pupil area of the subject when viewing the non-target screen; and the measurement and a test unit for statistical test the difference between the target screen pupil area measured with the non-target screen pupil size by means, said presenting means in advance for each subject Based on the findings on the familiarity made, closeness is selected more strings threshold, and wherein the string and be Rukoto to the appropriate string the selected in the predetermined range.

The preference determination system according to claim 3 is the invention according to claim 1 or 2 , wherein the presenting means alternately displays the target screen and the non-target screen at the center of the screen, and displays the non-target screen. In this case, a character string in which only several “+” s are displayed in white color on a black background screen is displayed, and when the target screen is displayed, the display order is randomly changed.

The preference determination program according to claim 4 is at least one or more character strings included in a character string group that is a collection of character strings related to preference, and calculates the familiarity of the character strings in advance to a certain range. Presenting the target screen with the corresponding character string and the non-target screen that is a blank without the character string or the character string not included in the character string group to the subject, and the subject when viewing the target screen The target screen pupil area that is the pupil area and the non-target screen pupil area that is the pupil area of the subject when viewing the non-target screen, and the target screen pupil area and the non-target screen measured by the measuring means When statistically testing the difference with the pupil area and presenting the difference, the familiarity is calculated based on the results of the investigation on the familiarity conducted for each subject in advance. Select a string of more value to execute a string the selected a computer to a character string corresponding to the predetermined range.

  According to this method, by controlling the word familiarity to a certain level or more, the influence of cognitive mental activity on the pupil area can be suppressed, and the preference determination can be realized with high accuracy.

The preference determination program according to claim 5 is at least one or more character strings included in a character string group that is a collection of character strings related to preference, and calculates the familiarity of the character strings in advance to a certain range. Presenting a target screen with a corresponding character string and a non-target screen with a character string not included or not included in the character string group to a subject in a range of 0.5 seconds to less than 5 seconds, the target screen The target screen pupil area that is the pupil area of the subject when viewing the screen and the non-target screen pupil area that is the pupil area of the subject when viewing the non-target screen, and the target measured by the measuring means and statistical test the difference between the the screen pupillary non-target screen pupil size, when the presentation is findings on the degree of intimacy that is performed in advance for each subject Based on, familiarity is selected more strings threshold, to execute a string the selected a computer to a character string corresponding to the predetermined range.

  The present invention can provide a preference determination method with little room for deception.

The example of the procedure of the preference determination method of this embodiment is shown. The example of the character string extracted beforehand shown on the target screen of the preference determination method of this embodiment is shown. The example of the target screen and non-target screen of the preference determination method of this embodiment is shown. The example of the measurement result of the pupil area with respect to the character string regarding the occupation which concerns on the preference determination method of this embodiment is shown. It shows about the verification result of the preference determination method of this embodiment.

  The present invention will be described with reference to the accompanying drawings. Embodiment described below is an example of implementation of this invention, and this invention is not restrict | limited to the following. In the present specification and drawings, the same reference numerals denote the same components.

  First, with reference to FIG. 1, the preference determination method of this embodiment is demonstrated. FIG. 1 is an example of a procedure according to the preference determination method of the present embodiment. As shown in FIG. 1, the preference determination method of the present embodiment is included in a target screen and a character string group in which at least one or more character strings are included in a character string group that is a collection of character strings related to preference. When viewing the target screen pupil area and the non-target screen, which is the pupil area of the subject when the target screen is presented to the subject and the target screen is displayed as a blank non-target screen with a non-printed character string The non-target screen pupil area, which is the pupil area of the subject, is measured, and the difference between the target screen pupil area and the non-target screen pupil area is statistically tested.

  A procedure for executing a specific preference determination method of the present embodiment will be described. In carrying out this embodiment, first, the subject is positioned in front of the monitor screen to prepare a system for observing the pupil of the subject. In S101, a non-target screen is presented to the subject, and in S102, the non-target screen pupil area is measured when the subject is viewing the non-target screen. For example, a plurality of “+” is displayed on the non-target screen. Next, the target screen is presented to the subject in S103, and the target screen pupil area of the subject is measured when the subject is looking at the target screen in S104. The character string displayed on the target screen includes not only characters but also symbols. In step S105, the procedure is repeated until the presentation of all target screens is completed. It is also possible to cause the computer to execute S101 to S105 as a preference determination program, display a non-target screen or a target screen on the monitor screen, and control a pupil area measuring device described later to measure the pupil area. Here, the procedure for displaying the non-target screen ahead of the target screen has been described, but the non-target screen may be displayed after the target screen.

  In the present embodiment, a character string related to the preference is shown to the subject, and the preference is digitized using the pupil area of the subject as an index to determine the presence and strength of the preference. Specifically, this embodiment includes a target screen on which a character string related to preference is displayed and a non-target screen on which a character string not related to preference is displayed or a blank screen is displayed. There are as many target screens as the number of character strings. Further, only one character string is displayed on each target screen. For example, character strings related to different occupations such as “biologist” and “movie director” are displayed on different target screens, and two target screens are prepared. There may be at least one target screen. Then, the non-target screen and the target screen are alternately displayed on the subject, and the subject when viewing each target screen with the pupil area of the subject when viewing the non-target screen as the non-target screen pupil area. Each target screen pupil area is measured with the pupil area of the target screen pupil area as the target screen pupil area. Finally, a target that compares the non-target screen pupil area average value with each target screen pupil area average value and shows a larger target screen pupil area average value with statistical significance than the non-target screen pupil area average value. It can be determined that the subject's preference exists for the character string written on the screen. Not only the character string is related to the occupation, but also the preference other than the occupation may be determined with respect to the subject's preference using the pupil area as an index and only the character string as a presentation.

  The essence in the present embodiment is to use only a character string as a presentation, not a moving image, a person, or a landscape image when determining the preference. In the case of only a character string, unlike a moving image or a person or landscape image composed of pixels having various luminances, the feature is characterized in that the shape is visually recognized by a subject by pixels having the same luminance. For example, it is assumed that character strings related to the occupations “movie director” and “biologist” are displayed at different timings to determine the preference for occupation. The luminance of the character string “Director” and the character string “biologist” are almost the same unless the color or size of either character string is changed. On the other hand, for example, the respective luminances of two moving images of a news program and a sports program usually have a large difference between the luminances. The control of luminance is very easy when a character string is used, which is a great advantage of using a character string. By using only the character string as a presentation, it is possible to determine human preference based on the pupil area with almost no influence of luminance.

  There has never been an example in which only the character string is presented and the human preference is determined using the pupil area. The non-patent document 7 tries to measure the mental load when the subject is memorized only by using a numerical string as a presentation, but the memory is a mental activity related to “knowledge” in the cognitive domain, Since preference is a mental activity in the emotional region, the mental activity determined by the pupil area is different.

  Moreover, the preference determination method of this embodiment makes the brightness | luminance of a non-target screen when the brightness | luminance of a target screen and a character string are described into a fixed range. By setting the luminance of each target screen displaying the target character string within a certain range, the influence of the luminance of the screen can be eliminated. The pupil is greatly affected by the brightness of the screen, but the brightness that varies depending on the number of characters and the number of strokes of the character can be adjusted by the brightness of the character, the thickness of the line, or the size of the character. Also, the brightness of the non-target screen can be measured with high accuracy by using the same brightness as that of the target screen. That is, by using a character string as a presentation, it becomes easy to control the brightness between the screens. As a result, the influence of luminance on the change in pupil area can be minimized. In addition, after the preference determination method of the present embodiment is executed by a computer as a preference determination program and a target screen or a non-target screen is displayed in advance on a monitor screen and the luminance is measured, the character is so determined that the luminance is in a certain range. The brightness, line thickness, or character size may be adjusted.

  Here, with reference to FIG. 2 to FIG. 4, four processes for carrying out the present embodiment will be described. The first process is extraction of a character string to be presented from a prepared character string. The second process is presentation of a character string. The third process is measurement. The fourth process is data using a program. It is preference determination by analysis. The preference determination method and program of this embodiment are characterized by these four processes.

  First, referring to FIG. 2, an example relating to extraction of a character string to be presented on the target screen is shown as a first process. It is preferable that the experimenter who measures the preference prepares a character string group in advance and then satisfies the following conditions when extracting the character string to be presented. In other words, the number of mora of the character string is extracted within a certain range and the familiarity of the character string is extracted within a certain range. In the preference determination method of the present embodiment, the number of mora or familiarity of the character string is calculated in advance for the character string written on the target screen, and only the character string corresponding to a certain range is written on the target screen.

  The purpose of extracting and extracting the number of mora of a character string within a certain range in advance is to control the luminance of the character string to be presented between character strings. As described above, the character string has a great advantage in controlling the luminance. However, even when the character string is presented on the screen, the luminance between the screens is different if the number of mora of the character string is greatly different. Make a big difference. For example, if the strings “High School Teacher” and “Market Development Consultant” are presented in white letters on a black background, the one with the larger number of mora in the string, that is, the “Market Development Consultant” is white on the screen. The area increases. The screen brightness increases as the white area increases. An increase in luminance leads to a reduction in the pupil area, which affects the preference determination based on the pupil area. Therefore, it is preferable to control the luminance between target screens within a certain range by controlling the number of mora of the character string within a certain range. In this embodiment, the preference determination of the subject can be realized by controlling the number of mora of the character string within a certain range. The number of mora is defined as the number of mora, that is, the number of kana characters in a word. For example, a character string in which the number of mora falls within a certain range may be extracted from a character string prepared in advance by causing the computer to execute the preference determination method of the present embodiment as a preference determination program.

  The purpose of extracting the intimacy of the character string within a certain range in advance is to focus on the mental activity of preference. A character string with a low familiarity is a character string with a low degree of familiarity. Therefore, when such a character string is presented in front of the eyes, there is a possibility that the subject will have a mental burden for understanding. Although the purpose of this embodiment is not mental activity in the cognitive domain of understanding but preference, as already stated, if the intimacy of the string to be presented is not controlled, the mental load is imposed by the string. Invite states that may or may not occur. It has been confirmed in Non-Patent Document 7 that the mental load caused by mental activity in the cognitive area affects the pupil area. The mental burden for understanding lowers the accuracy of preference determination. Therefore, it is preferable to control the mental load within a certain range by controlling the familiarity of the character string within a certain range, so that the screen on which the character string is displayed, that is, the target screen is displayed. In the present embodiment, the preference determination of the subject can be realized by controlling the closeness of the character string within a certain range. The familiarity is defined as an index representing how familiar a certain character string is felt. For example, the familiarity with respect to a character string may be acquired in advance, and the preference determination method of the present embodiment may be executed by a computer as a preference determination program to extract a character string with a familiarity within a certain range.

  The character string shown in FIG. 2 is a character string related to a profession having a character string with a mora number of 5 to 10 and high intimacy. The number of mora was calculated based on the list of occupations existing in the society created by the inventors' investigation. In addition, regarding the character strings of occupations satisfying the condition of the number of mora, the inventor investigated how close the five graduate students were to closeness, and selected character strings with a high level of closeness. Specifically, the degree of intimacy was a seven-step method from 1 to 7, with 7 being the highest and 1 being the lowest. As a result of instructing graduate students as subjects to rate each character string related to occupation, an average of 6.0 or more The string that showed the value was defined as a profession with a high degree of familiarity. Note that FIG. 2 is an example, and a character string different from the example in FIG. 2 is defined according to the strategy of the practitioner of the present embodiment, that is, the prepared character string group, and the condition setting of the number of mora and the familiarity. Sometimes.

  Next, with reference to FIG. 3, an example of a non-target screen and a target screen in the process of presenting a character string is shown as the second process. 3A is a non-target screen, FIG. 3B is a target screen, FIG. 3C is a non-target screen, FIG. 3D is a target screen, and FIG. ) Shows a non-target screen, and FIG. 3F shows a target screen. “+” Of the non-target screen in FIG. 3 represents a character of a character string that is not included in the character string group that is a collection of character strings related to the preference written on the target screen, and includes a symbol. In the process of presenting the character string shown in FIG. 3, it is desirable to have the following display operation. That is, alternately display a target screen and a non-target screen, display each screen preferably for about 0.5 seconds or more and less than 5 seconds, and display a character string in a size and color equal to the center of the screen and on a black background screen. It is to display in white color and not to give a cognitive task.

  The purpose of alternately displaying the non-target screen and the target screen is to eliminate the residual effect. The residual effect means that the pupil area response to the target screen previously presented in time series affects the pupil area response to another target screen presented immediately after. For example, after presenting the target screen for the character string “movie director” and immediately presenting the target screen for the character string “biologist”, the short-term memory of the former target screen is stored in the pupil area for the latter target screen. May have an effect, and it is difficult to ensure correspondence between the pupil area and each target screen. In this embodiment, measurement of pupil area corresponding to each target screen is realized by inserting a non-target screen each time between the target screen presented before and another target screen presented after that. To do.

  The presentation time of the non-target screen and the target screen will be described. The presentation time of each screen is preferably about 0.5 seconds or more and less than 5 seconds. It is physiologically confirmed that it takes about 0.5 seconds for the reaction of pupil enlargement or reduction to occur after an object is presented in front of the eyes. Also, if you look at the same presentation for more than 5 seconds, you may get bored or tired, and factors such as stress may affect the pupil area. is there. In the present embodiment, the measurement time of the pupil area response due to the preference for each presented screen by setting the presentation time of each target screen or each non-target screen to preferably about 0.5 seconds or more and less than 5 seconds. Can be realized.

  The purpose of displaying the character string in the same size and color at the center of the screen and in white color on the black background screen is to unify the screen brightness and measure the pupil response with high accuracy. Since the size and color of character strings affect screen brightness, it is preferable to unify them. In addition, if the presentation position of the character string on the target screen is not uniform between the target screens, there will be a difference in the visual condition, so that not only the factor of preference but also the factor of visual condition may affect the pupil area It is preferable to always display the character string at the center of the screen. And if the character string is displayed with white background and black characters on the screen, it is expected that the area occupied by the entire white background screen will normally be larger than the area occupied by the black character overall screen. Overall brightness increases overall. As described above, when the luminance is high, the pupil shrinks. However, in the case of the white background as described above, the pupil shows a contraction reaction as a whole regardless of the preference of the subject, and the pupil It may be difficult to obtain a change reaction. Therefore, it is preferable to display the target screen with white characters on the black background screen in order to keep the brightness of the entire screen low and to easily obtain a pupil change reaction. For example, only a few white color “+” s are displayed on the black background screen. In the present embodiment, by displaying the character string in the same size and color at the center of the screen and in white color on the black background screen, it is possible to measure the pupil area change due to preference.

  The purpose of not giving the cognitive task is to narrow down the factors affecting the pupil area to the mental activity of preference. A cognitive task is a task related to mental activity in the cognitive domain such as memory and calculation of character strings. By giving a cognitive task to a subject, the pupil area responds not only to the mental activity of preference but also to cognitive mental activity. May produce results. In the present embodiment, while the subject is viewing the target screen on which the character string is presented and the non-target screen, the recognition task is not given, and the subject is only asked to view the presented screen. Focus on the preference, which is a spontaneous or intrinsic emotion, and realize the presence / absence or strength determination.

  Further, the measurement process as the third process preferably has the following measurement conditions. In other words, using a hardware such as a web camera with an infrared irradiation function or a commercially available dedicated hardware, the pupil area is measured according to a method such as the corneal reflection method, and brightness is controlled in order to control brightness. It is to measure at a place where the natural light of the changing sun is blocked. For the measurement of the pupil area, a known method can be applied using a camera with an infrared irradiation function or the like.

  The process of preference determination by data analysis, which is the fourth process, preferably includes the following operations. In other words, the target screen pupil area is normalized by the normal pupil area that is different for each subject, the target screen pupil area is the average value of the pupil area while the subject is viewing each target screen, and the non-target screen The pupil area is an average value of the pupil area while the subject is viewing the non-target screen, correcting pupil area data at the time of blinking, and testing a statistically significant difference by comparison within an individual. You may make a computer perform the preference determination method of this embodiment as a preference determination program.

  The purpose of standardizing data for each subject is to control factors of individual differences in pupil area. It is physiologically known that the pupil area varies from individual to individual. For example, when a subject's normal pupil diameter is 3 mm and another subject's normal pupil diameter is 5 mm, and each subject is presented with the character string “High School Teacher” as the target screen, It is assumed that the pupil diameter when viewing the row is 4 mm and the length of B is also 4 mm. In the former, the pupil diameter increased, while in the second, the pupil diameter decreased. This means that when a pupil diameter of 4 mm is measured, it cannot be generally determined that there is preference for the character string regardless of the subject. It is desirable to standardize the data in consideration of factors of individual differences. Specifically, the normalization means that the pupil area during visual recognition of the character string, that is, the target screen pupil area, is divided by the normal pupil area instead of the absolute value of the subject's pupil area, that is, the measured value as it is. Is to capture the value as a value after normalization. For the previous target screen example “High School Teacher”, the pupil area value of the former was an absolute value, that is, a measured value of 4 mm, but the value after normalization was 4/3 = about 1.33, after the standardization of B The value of 4/5 = 0.8. It can be concluded that, due to the standardization, the pupil diameter was relatively increased with respect to the character string of “High School Teacher” and that the pupil diameter was relatively decreased with respect to the character string. The normal pupil area is the pupil area while viewing the non-target screen, that is, the non-target screen pupil area. In addition, since the pupil is substantially circular, the procedure according to the present embodiment is not affected even if the notation or calculation target is the pupil diameter or the pupil area. The present embodiment is characterized in that by performing normalization, it is possible to absorb the individual difference factor of the natural pupil area and to determine the preference of each subject.

  The average value of the pupil area for each screen is obtained by obtaining the typical pupil area of the subject during the time when the non-target screen is presented and the representative value of the target screen pupil area for the presented character string. Objective. The normal pupil area is the pupil area when the non-target screen is presented. When the non-target screen is displayed for 2 seconds, for example, it is represented by the average value of the pupil area data measured for 2 seconds. For example, when the pupil area is measured at 60 Hz, it is represented by an average value of pupil area data of 120 frames. This is equivalent to the average value of the non-target screen pupil area already described. Also, when each character string is presented, the target screen pupil area average value of each character string is calculated in the same procedure, and is taken as a representative value of the pupil area value for each character string. As mentioned earlier, there is a latency of about 0.5 seconds after the presentation of the object in the pupil reaction, so it is also possible to consider a method of excluding pupil area data about 0.5 seconds after the presentation of the object from each average value calculation target. It is done. In addition, here, the data for which the average value is calculated in the target screen pupil area is not the measured value as it is, but the normalized value described above is desirable. The present embodiment is characterized in that the average value of the non-target screen pupil area and the average value of the target screen pupil area can be calculated so that the normal pupil area and the representative value of the pupil area for each character string can be grasped.

  The purpose of correcting the pupil area data at the time of blinking is to capture an approximate value of the true value of the pupil area. Due to the structure of the eyes, the pupil is covered by a wing during blinking, and the pupil area cannot be measured because infrared rays do not reach the pupil temporarily. In that case, the measured pupil area data is recorded as zero or an error value. In fact, it is not physiologically possible that the pupil area becomes zero, but it is simply a data loss resulting from being obscured by the eyelids. If such missing data is recorded as zero, it will have a large effect on the calculation of the average pupil area for each screen, specifically, the more zero time zone, that is, the blink time zone, The average pupil area of the screen becomes smaller due to the influence of zero. In this case, it becomes difficult to grasp the pupil area response due to the preference factor. Therefore, it is desirable to correct the pupil area at the time of blinking. Specifically, a correction method using a substitute value after estimating the pupil area at the time of blinking by a moving average of pupil areas before and after blinking, a correction method using a support vector regression model, and the like are conceivable. This embodiment is characterized in that by correcting the pupil area at the time of blinking, the influence of the blinking on the average pupil area value for each screen can be reduced, and preference determination can be realized.

  In addition, in relation to the pupil area at the time of blinking, there is a method that does not generate blink data by instructing the subject to suppress blinking while presenting the screen. Eye fatigue and mental load resulting from suppression can affect pupil area.

  The purpose of testing a statistically significant difference by intra-personal comparison is to determine a character string having preference. A non-target screen pupil area average value and a target screen pupil area average value are each regarded as sample data, and then a statistical test is performed to determine whether there is a difference between the two average values. For example, by performing a statistical test on the difference between the average value of the target screen pupil area corresponding to the target screen of “high school teacher” and the average value of the non-target screen pupil area, the profession of “high school teacher” The presence / absence and strength of palatability are determined. Specifically, after statistically testing whether the two data are equal variances, if they are equal variances, a normal t-test is performed. If they are not equal variances, t A test is performed to statistically test whether there is a difference between the two average values. If each sample data is not normally distributed, a non-parametric test is performed. When statistical significance is recognized as a result of the test, it is determined that there is a preference if the average value of the target screen pupil area is larger than the average value of the non-target screen pupil area, and it is determined that there is no preference if it is small. Moreover, the magnitude matches the degree of preference. The target screen pupil area average value and the non-target screen pupil area average value are compared with values of the same subject. For example, the average value of the target screen pupil area of the subject A and the average value of the non-target screen pupil area of the subject B are not compared, and an intra-personal comparison is made. For example, the preference determination method of the present embodiment may be executed by a computer as a preference determination program to perform a statistical test, and a character string or a statistical test result may be displayed on the monitor screen.

  In addition, when the obtained pupil area data is regarded as sample data, it is preferable to perform a statistical test by the above method. However, when it is regarded as all data, a simple difference calculation is performed regardless of the above method. You may go.

  With reference to FIG. 4, an example of the determination result of the preference for occupation in subject A is shown. FIG. 4A shows the non-target screen pupil area average value. The horizontal axis indicates each character string relating to occupation, and the vertical axis indicates the average value of the target screen pupil area after normalization. And as a result of having performed the statistical test of the difference of the average value about each target screen pupil area average value and non-target screen pupil area average value by the procedure mentioned above, statistical significance difference was recognized by 5% significance level. Each character string is marked with “*” as shown in FIG. The pupil area measurement method, the data analysis method, the definition of the non-target screen pupil area average value, and the normalized target screen pupil area average value are as described above. Subject A is determined to have a preference for “humor writer”, “movie director”, and “illustrator”. On the other hand, it is determined that the “school clerk”, “cash register”, and “sales clerk” have no preference. It is determined that there is no emotion in the other character strings. Subject A does not want to get a job with a lot of routine work, but may have a desire to get a job with atypical work, especially a job that strongly demands creativity. It is an example of interpretation of a result. Since this result changes according to individual preference, it is predicted that the result will change if the subject changes.

  Note that any one of these preference determination methods can be realized as a program executed by a computer.

  Next, a verification result of the preference determination method of the present embodiment will be described with reference to FIG. FIG. 5 shows the result of an experiment conducted to verify the relationship between the pupil area response of the subject to the character string on each target screen and the preference of the subject. In FIG. 5, the vertical axis represents the normalized target screen pupil area average value, the horizontal axis represents the standardized target screen pupil area average value for the character string group that has a preference, and there is no preference. Two average values of the target screen pupil area average values after normalization for the answered character string group are shown. For each of these average values, after calculating the pupil area average value for each target screen, the average value data is classified according to the presence or absence of preference, and then the average value is calculated for each presence or absence of preference. Stepping. Then, a statistical test of equal variance is performed on the difference between the two average values, and after confirming that the variance is equal by two-sided test: F (41,25) = 1.912, p = 0.1246 A normal t-test was performed, and two-sided test: t (66) = − 2.0552, p <0.05 was obtained. In other words, there is a difference between the two average values at the 5% significance level.

  In the experiment in FIG. 5, first, a target screen on which each character string included in the character string group shown in FIG. 2 was presented to four subjects, and the pupil area was measured by the method described above. . Next, the preference about each character string contained in the character string group shown in FIG. 2 was confirmed verbally with respect to the same four test subjects as the subjects who measured the pupil area. A statistical test was then performed on the relationship between pupil area and verbal preference. As a result, as shown in FIG. 5, the pupil area is enlarged for the character string group that is answered verbally and the pupil area is reduced for the character string group that is answered verbally is not preferred. It was confirmed. Character groups that the average value of the target screen pupil area after normalization for the character strings that are recognized as having a difference between the two average values at the 5% level and that there is preference is not preferred Since it exceeds the average value of the target screen pupil area after normalization, it was found that the preference determination method of the subject can be realized by the preference determination method according to the present embodiment.

  Here, four points are supplemented for the experimental results shown in FIG. The first point is recognition control. The second point is to confirm general trends. The third point is the interpretation of deception. The fourth point is order effect control.

  Explain the control of the first point of recognition. Regarding the experimental results shown in FIG. 5, it is not appropriate for the subject to confirm the preference for a character string whose image does not boil. This is because there is no preference for things that you do not know. Therefore, verbal questions were asked to four test subjects whether they knew the degree of recognition of each character string shown in FIG. 2, that is, the occupation indicated by each character string. As a result, there was a result that one subject A, four subject Bs, two subject nieces 2, and subject D did not know about five character strings. Therefore, 19 character groups, 16 subjects, 16 subjects, 18 subjects, and 15 subjects had 15 character strings as analysis targets in the experiment shown in FIG.

  Explain that the purpose is to confirm the second general tendency. The experimental results shown in FIG. 5 are not intended to be based on an individual, that is, a subject, but to verify the general relationship between the pupil area and the verbal preference, so the results are not analyzed for each subject. It is preferable to analyze four subjects collectively. Therefore, the character string group confirmed verbally when there is a preference and the character group group confirmed verbally when there is no preference are separated, and the average value of the target screen pupil area after each standardization is added to Divided by the number of columns. Thereby, the average pupil area value of the character string group confirmed verbally if there is a preference and the average pupil area value of the character string group confirmed verbally if there is no preference. When there is a preference, the number of character strings included in the character string group confirmed verbally is 26, and when there is no preference, the number of character strings included in the character string group confirmed verbally is 42. This total is 68, which matches the sum of the number of character strings to be analyzed for the above-mentioned subject A, B, B, and Ding.

  For example, in the case of the character string “school clerk” shown in FIG. 2, assuming that the subject A verbally responds that there is a preference and the subject B verbally answers that there is no preference, the character “school clerk” The average value of the target screen pupil area after the normalization of the subject A with respect to the column is classified into the “preferred” character string group shown in FIG. 5, and the average value of the target screen pupil area after the normalization of the subject B with respect to the character string is shown in FIG. 5 is classified into a character string group of “no preference” shown in FIG. Such processing is performed one by one for 26 “preferred” shown in FIG. 5 and 42 “not preferred” shown in FIG. In the confirmation of verbal preference, there were three choices of “with preference”, “no preference”, and “cannot say either”, but there was no corresponding to “cannot say either”.

  All four subjects are graduate students who can read Japanese, have a willingness to find a job, have determined their desired job type, and have normal vision. There is no individual difference about them. Control of individual differences among survey subjects suitable for confirming general trends.

  The interpretation of the third point of deception will be explained. As already mentioned, in the conventional inspection method, the interview inspection cannot completely eliminate the deception of the answer of the job applicant. From this problem, it may be pointed out that the subject may give a deceptive answer because the test shown in FIG. However, the experiment shown in FIG. 5 has a completely different environment from the scene of the employment test in a company or the like. In other words, in recruitment interviews at companies, etc., those who wish to work, such as students who are confirmed verbally in their vocational preferences, are motivated to pretend to be highly evaluated by companies. In the experiment shown in 5, there is no reward that motivates to pose an answer. The experiment shown in FIG. 5 is not a scene of recruitment selection, and no reward is prepared for the subject such as employment placement for the subject and setting a reward amount according to the answer result. Therefore, the oral confirmation result on the presence or absence of preference in FIG. 5 is reliable.

  The fourth point order effect control will be described. If the character strings are presented to all four subjects in the same order, there is a possibility that the subject will have a specific meaning. This is a case where the subject recalls a specific image according to the order of the character strings. Since such image recall should be excluded for the purpose of the experiment shown in FIG. 5 for determining the preference for each character string, the order of the character strings presented for each subject was randomized.

  As described above, as shown in FIG. 5, it has been confirmed that the preference of the subject can be realized by the method according to the present embodiment. Note that the experimental results shown in FIG. 5 are only for verification of the effect of the present embodiment, and the same process as these experiments may not be performed when the present embodiment is performed.

  As described above, in the preference determination method of the present embodiment, the character string displayed on the target screen may be a character string related to occupation. Since the deception property is likely to be mixed in the occupational preference determination scene, the occupational preference is a preference suitable for applying this embodiment.

  Regarding the determination of occupational preferences, deceptiveness is likely to be mixed in the conventional method because psychological reward expectations exist. In particular, in the current situation where the job offer ratio is low, job applicants do not necessarily have a strong preference for the companies that are taking the recruitment examination, but the purpose is to first find a job at any company. It can be easily assumed to take the action. For example, in questionnaire surveys, it is assumed that job applicants will consider strategies that pretend to be desirable professions to be highly evaluated by companies that are the subject of the inspection, regarding the answers to each question in the inspection. . Also in interview examinations, it is assumed that those who wish to work give a desirable response in order to be highly evaluated by the company, etc., and give explanations about their preference for occupation. There are already many books that discuss trends and countermeasures for recruitment interviews and questionnaires used in recruitment screening, and rather than answers or responses based on the true vocational preference of job seekers. It is also easily assumed when a job applicant examines an answer or response strategy for passing the employment examination of a company or the like.

  In order to clarify the problem of the questionnaire survey in more detail, the inventors conducted an experiment using the Japanese version of the VPI occupational interest test. An experimental group and a control group were established, and 10 students were used as subjects in each group, and the subjects were different among the groups. Both groups took the Japanese version of the VPI Vocational Interest Test twice. In the experimental group, the subject assumed the scene of the employment examination, and the subject was instructed to answer when the type of application was different between the first examination and the second examination. There was no such instruction in the control group, and it encouraged subjects to give an honest answer. In each group, the average of the first test result and the average of the second test result were compared for each scale, and the difference between the averages was confirmed by t-test (5% significance level). The Japanese version of the VPI Occupational Interest Test is composed of 6 scales, and the occupational interest of the subject is determined by the score of each scale. As a result, 6 groups out of 6 scales (R, I, A, S, E, C) were significantly different in the experimental group, whereas no significant difference was observed in all scales in the control group. It was. In other words, in the experimental group where the subject is expected to receive a psychological reward of passing the employment examination, the subject's response tendency changes significantly, and in the control group where no reward is assumed, the subject's response tendency does not change I understood. Since the employment examination is applicable when there is a psychological reward for the job applicant, the Japanese version of the VPI Vocational Interest Test may result in a deceiving answer from the job applicant, and the preference for a true occupation cannot be determined. It was confirmed.

  The invention described in the present embodiment is used, for example, in recruitment activities in organizations such as companies or government agencies that determine job preferences of job applicants. Or, it is also used by organizations such as companies or administrative organizations that conduct employee or staff transfer and personnel assessments based on their preference for occupation. It is also used in educational institutions such as schools and cram schools that provide career counseling for students based on their preference for occupation. It is also used by recruitment agencies and temporary staffing companies that provide career counseling for those who wish to work or change their jobs, based on their preference for occupation. Alternatively, it is also used in consulting companies and general research institutes that provide consulting based on their preference for occupation.

  In the preference determination method of the present embodiment, the character string displayed on the target screen may be a character string related to a certain product or service, a certain teaching material in a school, or the like. In a market survey of products and services in a company, consumer preferences are determined, and in educational institutions such as schools, the interests of students and students are determined. Those preferences are preferences suitable for applying this embodiment.

  In the present embodiment, the occupation has been described as an example of the preference, but the present invention is not limited to the occupation and can be applied to the following examples. For example, when making announcements and negotiations, the audience and negotiating partners are the subjects, and the materials used in the announcements and negotiations are presented, and the preference for the presentations is determined, attracting the audience's interest, and prompting the buying behavior of the negotiating partners. There are cases where the purpose is to do. Or at a school or a cram school, using students as subjects, classes and teaching materials as presentations, determining preferences for the presentations, attracting students' interests, and evaluating the quality of classes and teaching materials It can also be used for purposes. In addition, it may be aimed at investigating investor subjects, stock prices, etc. as presentations, determining preferences for the presentations, and investigating the decision-making timing and decision-making structure of investment behavior in specific securities, etc. Conceivable.

  Since the preference determination method and the preference determination program of the present invention can determine a subject's preference, they can be used for various occupational aptitude surveys and market surveys.

S101: Non-target screen presentation S102: Measure pupil area at non-target screen presentation S103: Target screen presentation S104: Measure pupil area at target screen presentation S105: End determination of target screen presentation

Claims (5)

A target screen which is at least one or more character strings included in a character string group which is a collection of character strings related to preferences, and in which a character string corresponding to a certain range is expressed by calculating the familiarity of the character strings in advance Presenting means for presenting to the subject a non-target screen that is a blank that is not included in the character string group or that is not represented, and a target screen pupil that is the pupil area of the subject when viewing the target screen Measuring means for measuring a non-target screen pupil area, which is a pupil area of a subject when viewing the non-target screen, and the target screen pupil area measured by the measurement means and the non-target screen pupil area A test means for statistically testing the difference,
The presenting means selects a character string whose familiarity is greater than or equal to a threshold value based on an investigation result regarding the familiarity previously performed for each subject, and sets the selected character string as a character string corresponding to the certain range. A taste determination system characterized by this. A target screen which is at least one or more character strings included in a character string group which is a collection of character strings related to preferences, and in which a character string corresponding to a certain range is expressed by calculating the familiarity of the character strings in advance Presenting means for presenting a non-target screen in which a character string not included in the character string group is described or a blank space without description to a subject in a range of 0.5 seconds to less than 5 seconds, and a subject when viewing the target screen Measuring means for measuring the target screen pupil area that is the pupil area of the subject and the non-target screen pupil area that is the pupil area of the subject when viewing the non-target screen, and the target screen pupil area measured by the measuring means, wherein a test means for statistical test the difference between the non-target screen pupil size, the presenting means, research focusing relating to the degree of intimacy that is performed in advance for each subject Preference determination system, closeness is selected more strings threshold, characterized in strings and to Rukoto to the appropriate string the selected in the predetermined range based on. The presenting means alternately displays the target screen and the non-target screen in the center of the screen, and when displaying the non-target screen, displays only a few “+” in white color on the black background screen. The preference determination system according to claim 1 or 2 , wherein when the target character screen is displayed and the target screen is displayed, the display order is randomly changed.   A target screen which is at least one or more character strings included in a character string group which is a collection of character strings related to preferences, and in which a character string corresponding to a certain range is expressed by calculating the familiarity of the character strings in advance The target screen pupil area that is the pupil area of the subject when viewing the target screen by presenting the subject with a non-target screen that is not included in the character string group or that is a blank without the notation, and the target screen pupil area Measure the non-target screen pupil area, which is the pupil area of the subject when viewing the non-target screen, and statistically test the difference between the target screen pupil area measured by the measuring means and the non-target screen pupil area When presenting, a character string having a closeness greater than or equal to a threshold value is selected based on a result of investigation on the closeness previously performed for each subject, and the selection is performed. Preference determination program for executing the string to a character string corresponding to the predetermined range to the computer. A target screen which is at least one or more character strings included in a character string group which is a collection of character strings related to preferences, and in which a character string corresponding to a certain range is expressed by calculating the familiarity of the character strings in advance The subject's pupillary area when a non-target screen in which a character string that is not included in the character string group is written or a blank space that is not written is presented to the subject for 0.5 seconds to less than 5 seconds and the target screen is viewed The target screen pupil area and the non-target screen pupil area, which is the pupil area of the subject when viewing the non-target screen, and the target screen pupil area and the non-target screen pupil area measured by the measuring means and statistical test the difference between the time of the presentation, on the basis of the findings on the degree of intimacy that is performed in advance for each subject, familiarity is equal to or larger than the threshold string The selected preference determination program for executing the string the selected a computer to a character string corresponding to the predetermined range.

Images