Detailed Study of the Semantic Clusters of Extraverted Sensing (Socionic Function Se, “Force Sensing”)

V. L. Talanov, St. Petersburg, February 2016

Contents:

1. Abstract
2. Method
3. Semantic Clusters of Extraverted Sensing and Their Socionic Profiles
   • Type Profiles of the Clusters
   • Trait Profiles of the Clusters
   • Functional Profiles of the Clusters
   • Calculation of Functions from the Trait Profile
   • Profiles of the 15 Generalized Cluster Groups
4. Characteristics of the 15 Generalized Cluster Groups
5. Unity of the Extraverted Sensing Factor – Evidence from Direct Measurements and Analysis of Variance
6. Extraverted Sensing and Evolution
7. Briefly on Research into the Biological Basis of Se
8. On Methodological Shortcomings in the Psychological Component of Contemporary Neurogenetic and Neuroimaging Research
9. Extraverted Sensing and the Development of Society
10. Extraverted Sensing and Power
11. References

Abstract

In socionics, that is, a branch of neo-Jungian personality typology widespread in Russia and Ukraine, the temperamental level of human personality is classified on the basis of the hierarchy of eight basic psychic functions inherent in each person. Among these functions are two sensing functions – so-called “black” sensing (also “extraverted,” “static,” “force sensing,” or simply “Se”) and so-called “white” sensing (also “introverted,” “dynamic,” “sensing of sensations,” or simply “Si”). On the basis of the hierarchy of the 8 functions of the psyche, socionics distinguishes 16 psychological types, each of which possesses a set of psychological characteristics that is predominantly specific to it. The reverse approach is also possible, in which people are first classified into types on the basis of marker properties of the types (that is, into 16 conventionally distinguished sectors in a multidimensional psychological property space, with each person assigned to the one of the 16 sectors closest to them), after which, on the basis of this partition, the properties of the 8 psychic functions are studied. Using this latter approach, the author obtained reliable functional profiles characterizing 6663 questionnaire items, including both items used in socionics for psychotype diagnosis and “ballast” items. The resulting material made it possible to analyze the semantic clusters formed by questionnaire statements most closely associated with each of the 8 socionic functions of the psyche. The present work presents in detail the results of this analysis for extraverted sensing. It is shown that the so-called socially negative properties of extraverted sensing (aggression, egocentrism, a tendency toward sadism and profiteering, criminal tendencies, etc.), whose immanent inclusion among Se markers and, more broadly, among markers of the innate temperamental level is often disputed by socionists and psychologists, are indeed immanent properties of Se. Moreover, it is shown that the core of the numerous manifestations of socionic extraverted sensing consists precisely of those characteristics most closely associated with territorial aggression. These results are additionally confirmed by an independent correlational and analysis-of-variance study conducted outside socionic diagnosis, without using it at all. Analysis of variance also showed that environmental factors influencing the formation of the conditionally socially negative properties of extraverted sensing (upbringing, profession, etc.) are minimal: they account for a share of the variance in the magnitude of these properties five times smaller than the share associated with innate (although not necessarily hereditary) temperamental factors described by a person’s socionic profile. Finally, the article briefly discusses hypotheses concerning the physiological basis of Se, offers comments on the psychological support of contemporary neuroimaging and neurogenetic research, and analyzes the evolutionary origins of extraverted sensing as an independent function of the psyche and its relationships with society and power in different historical periods.

Method

For the purposes of the present study, the semantic clusters of extraverted sensing were analyzed by the author using 6663 initial questionnaire items with known socionic type, trait, and functional profiles (validated on samples ranging from 300 to 7000 people). The clusters were then assembled manually from approximately 2350 selected items with an experimentally identified extreme loading on Se relative to the other socionic functions. Items were grouped into clusters according to semantic similarity of their wording, while controlling that the type profile of the item under consideration correlated preferentially with the profiles of other items in that particular cluster. The resulting 122 semantic clusters of extraverted sensing were subsequently divided, with the aid of EXCEL, into 15 groups (superclusters), this time solely on the basis of similarity between their socionic profiles.

The initial type profiles of the 6663 items were obtained using several self-learning sociodiagnostic questionnaires. Each type profile consists of 16 numbers, one for each of the 16 socionic psychotypes. Each number is equal to the deviation from the population mean (in units of the standard deviation of the population) shown by representatives of the corresponding psychotype in the degree to which they agree with the given questionnaire statement. As a rule, several questionnaires in which the given item was tested are used to calculate each number. The final number is then the weighted mean of the results from all questionnaires, taking into account the number of representatives of the given type among their respondents. The 16 numbers constituting the type profile of a questionnaire item can be regarded as diagnostic coefficients for subsequently calculating the type profiles of new respondents answering this and other items in a subsequent questionnaire. A respondent’s complete type profile is formed from 16 linear correlation coefficients calculated between the vector of the respondent’s answers to all questionnaire items (after normalization) and 16 vectors of diagnostic coefficients, one for each of the 16 types, taken from the known type profiles of the corresponding questionnaire items.

At the beginning of the author’s work with sociodiagnostic questionnaires, namely several years ago at the start of the research program, the initial diagnostic coefficients of the questionnaires were obtained entirely, as a first approximation, from the sociotypes (one of the 16 possible types) declared in advance by the participants in the testing, also taking into account their declared probability and their declared socionics experience. On the basis of the coefficients obtained, first-stage type diagnosis was performed independently of the declared types, allowing the types of all respondents to be determined, both those who had declared their types in advance and those who had not. At this stage, declared psychotypes that differed grossly from the newly diagnosed ones were rejected, and the training procedure for obtaining diagnostic coefficients from the responses of respondents with previously declared psychotypes was repeated. Diagnosis was then performed again using the refined system of coefficients. The results of this diagnosis, now for all respondents in the sample, were again used to obtain a system of diagnostic coefficients for diagnosing each of the 16 psychotypes, but this was now done without regard to the declared types.

The diagnostic coefficients obtained were used in subsequent questionnaires containing the same items. Material accumulated from new questionnaires was used to refine the previously obtained system of diagnostic coefficients. At the final stage, when all material from all questionnaires had already been collected, an additional procedure for symmetrizing the diagnostic coefficients was performed. The point is that the system of diagnostic coefficients and the corresponding socionic type profiles of all respondents obtained with its aid deviate quite strongly from symmetry. This means that socionic traits that are supposed to be orthogonal (that is, to correlate with one another at a zero correlation coefficient) are in fact far from orthogonal in the full respondent dataset – even after correcting (balancing) the sample, by means of a special mathematical-statistical procedure, for inequality in the representation of the types within it. This departure from orthogonality, which reveals socionic asymmetry, affects both the basic (Jungian) traits and, to an even greater extent, the so-called 11 “weak” Reinin traits. For example, intuition is empirically found to correlate strongly with ethics and irrationality. Some of these “parasitic” correlations are justified and genuine because they have an understandable physiological basis, but some are indeed parasitic because they are caused by distortions in the psychotypes initially declared by respondents (for example, people prefer some psychotypes over others, and people’s self-identification by psychotype is also influenced by certain false but widespread socionic myths).

All these causes, reflecting both genuine physiologically determined asymmetry of socionic parameters and mere artifacts of human mentality, together produce distortions in socionic traits and cause their departure from mutual orthogonality. If this nonorthogonality is completely eliminated artificially in the system of diagnostic coefficients, the mean percentage of agreement between diagnosed and declared types falls from approximately 62% to 50%. However, the new subsequent systems of diagnostic coefficients obtained after this (on the basis of the types newly diagnosed with their aid) again gradually produce nonorthogonality of the traits. This should already be regarded as genuine asymmetry of socionic parameters, which repeatedly restores itself.

Using recurrent procedures and weighted criteria for optimal symmetrization (on the one hand maximizing the proportion of agreement between diagnosed and declared types, and on the other minimizing mutual correlations among the 15 socionic traits), the author succeeded in arriving at a system of diagnostic coefficients that, first, is stable under subsequent recurrent “self-learning” procedures applied to this system; second, provides minimal correlations among socionic traits; and third, despite all this, provides a proportion of agreement between the resulting type diagnoses and the types declared by respondents no worse than 58–59%. Thus, there are grounds to believe that this final system of diagnostic coefficients and the respondent type profiles obtained with its aid in any questionnaires best reflect the true structure of the distribution and interrelationships of socionic parameters in the population. For the purposes of the present work, the socionic profiles of questionnaire items obtained using precisely this system of diagnostic coefficients were used (it should immediately be noted, however, that using the “uncorrected” system of diagnostic coefficients, based only on respondents’ declared types, leads to exactly the same conclusions regarding the Se clusters).

Semantic Clusters of Extraverted Sensing and Their Socionic Profiles

The detailed item-by-item composition of all 122 semantic Se clusters distinguished by us can be found in a separate PDF file at the following address: https://yadi.sk/i/S-h8RbdjoxZcD

Here we provide only the list of clusters together with their socionic profiles. The type profile of each cluster is the result of averaging the type profiles of all items included in the cluster (taking into account, of course, the response version counted for the item: “yes” or “no”; for a “no” version, the profile added to the sum is multiplied by minus one). The standard deviation of the final cluster type profile is selected so that it equals the mean of the standard deviations of the type profiles of all items included in the cluster. The trait and functional profiles of a cluster are obtained from its type profile purely mathematically using the known procedures (described below in the presentation of Parts 2 and 3 of Table 1).

Table 1, Part 1. Type Profiles of the 122 Semantic Clusters of Extraverted Sensing. Clusters are combined into groups and presented in descending order of their exclusive association with extraverted sensing (see subsequent parts of Table 1). Each type profile is the result of averaging the type profiles of all items included in the cluster. In turn, for each individual questionnaire item in the cluster, the coefficients of its type profile show how far, on average, the property encoded in the item deviates from the population mean (in units of the standard deviation of the population) for representatives of each particular TIM among all 16. Type profiles form the basis for deriving all subsequent socionic profiles from them - including trait and functional profiles (see Table 1, Parts 2, 3, and 4)

Table 1, Part 2. Trait Profiles of the 122 Semantic Clusters of Extraverted Sensing. Trait profiles were obtained by formal calculation from the type profiles (the sum of the 8 type-profile coefficients for the TIMs at the positive pole of the trait is reduced by the sum of the 8 coefficients for the TIMs at the negative pole of the trait, and the result is divided by 16)

Table 1, Part 3. Decomposition of the 122 Semantic Clusters of Extraverted Sensing Across All 8 Socionic Functions. The values of the clusters for each function are the result of formally converting the cluster type profile (Table 1, Part 1) into a trait profile (Table 1, Part 2), and then into a functional profile using the coefficients in Table 2

Table 1, Part 4. Mean Cluster Magnitudes by Quadra. The quadra mean-TIM values are given in units of the standard deviation of the population and were recalculated from the type profile (to calculate a quadra mean, the type-profile coefficients corresponding to the TIMs belonging to the quadra are averaged)

When considering Table 1, Part 4, a question may arise: why do some extraverted sensing clusters statistically gravitate toward quadras for which Se is not a valued function? The point is that Se (not necessarily valued) is present in every TIM. For example, ESE and LSE have fairly strong Se, although their Se (in the strong “demonstrative” position) is not valued. Cluster no. 116 (thunderousness of the voice, loud laughter, guffawing), for example, predictably has its highest values for SLE and SEE, but it is also high for ESE and LSE (including not only because of Se, but also because of contributions to this cluster from Fe and Te, in addition to the principal contribution from Se). At the same time, introverts of the Se-valuing quadras have very low values for this cluster. Because of their low values, the quadra-mean values of this cluster for the second and third quadras are lower than the mean for Quadra 1.

A similar question may arise, for example, with respect to cluster no. 98 (predominance of consumer values): why does it gravitate mainly toward the third quadra, although the LIE sociotype from this quadra unquestionably does not share consumer values? The reason is that SEE makes the greatest contribution to Se in this quadra. Since SEE is also the “chief consumer” of the socion, it is precisely because of SEE’s contribution that, when the cluster values are averaged across all four types of the third quadra, this quadra takes the leading position among the socion’s quadras. Therefore, the table of quadra tendencies should not be absolutized – the tendencies shown in it are merely the result of averaging the contributions of four different types, and these tendencies do not necessarily extend to every type in the quadra under consideration. For information on a cluster’s association with each particular sociotype, one should consult not this table but Table 1, Part 1 directly, where the type profiles of the clusters are presented.

Table 2. Optimized Table of Coefficients for Calculating the 8 Functions from the Trait Profile. The value of a function is equal to the scalar product (the sum of element-wise products) of the function row in this table and the trait-profile row of the corresponding questionnaire item (or the trait profile of a generalized cluster from Table 1, Part 2, or a person’s trait profile). The resulting set of values for the 8 functions (that is, the functional profile) is already normalized to practically the same standard deviation as the corresponding type profile from which all calculations began. Functions obtained from traits using this system of coefficients subsequently provide exact inverse reconstruction of all traits associated with the functions as simple sums of the corresponding functions (without additional coefficients).

Table 3. Socionic Profiles of the 15 Generalized Clusters of Extraverted Sensing (Enlarged Groups Composed of the 122 Primary Clusters on the Basis of Profile Similarity)