1. Introduction
Understanding the personality traits of one’s self and others contributes to harmonious interpersonal relationships. However, getting to know one’s self and others in a short period of time is not an easy task, and inducing the personality traits of others is an even more difficult undertaking. In the Western world, studies on personality traits have a long and broad history [
1]. Many related studies have since followed, but the number of proposed personality characteristics has remained high. It was for this reason that Cattell [
2] converted these characteristics into 16 types of personality factor questionnaires. Later, Fiske [
3] performed a follow-up verification of Cattell’s research and derived the “Big Five” personality dimensions. In 1963, Norman [
4] verified Cattell’s procedures and announced that the five major factors constituted a reasonable method of personality classification.
Research on personality traits is core to many major disciplines, such as medicine, psychology and corporate management, whether for theoretical investigation or practical application [
5]. Personality traits stem from a consistent behavioral model and internal processes within each individual, allowing the individual to identify with a consistent behavioral model in different situations. The internal processes of personality traits include emotions, motivation and cognition. Although these processes occur at a deep level, they influence human behavior and feelings [
6]. Additionally, other studies have attempted to classify individuals into different personality types [
7]. For hundreds of years, the Chinese people have used physiognomy, palmistry (the ridges on the skin of the palm), bone reading and other methods related to physiological features to divulge an individual’s personality traits and fortune. To date, however, there are no studies that support a relationship between personality traits and fingerprints, which are an individually unique physiological feature.
Two features of fingerprints that are particularly important: (1) fingerprints do not change with time; and (2) every individual’s fingerprints are unique [
8]. Due to the above-described two characteristics, fingerprints have long been used for identification purposes [
9,
10]. Medina-Pérez proposed a new feature representation containing clockwise-arranged minutiae without a central minutia, a new similarity measure that shifted the triplets to find the best minutiae correspondence, and a global matching procedure that selected the alignment by maximizing the amount of global matching minutiae [
11]. In comparison with six verification algorithms, the proposed method achieved the highest accuracy in the lowest matching time. Ballan and Gurgen [
12] presented a method for fingerprint recognition based on principal component analysis and point patterns (minutiae) obtained from the directional histograms of a fingerprint. This study gave the same performance as that of the uncompressed data, but reduced computation. Yang et al. used fusion to enhance the biometric performance in template-protected biometric systems [
13]. They investigated several scenarios (multi-sample, multi-instance, multi-sensor, multi-algorithm and their combinations) on the binary decision level and evaluated the performance and fusion efficiency on a multi-sensor fingerprint database with 71,994 samples. Fingerprint image quality improvement was proposed in [
14]. The algorithm consists of two stages. The first stage is decomposing the input fingerprint image into four sub-bands by applying the two -dimensional discrete wavelet transform. At the second stage, the compensated image is produced by adaptively obtaining the compensation coefficient for each sub-band based on the referenced Gaussian template. The method concluded an improved clarity, quality and continuity of ridge structures, and therefore, the accuracy is also increased. Background and the blurred region of fingerprint images are also removed. Bartunek et al. [
15] presented several improvements to an adaptive fingerprint enhancement method that is based on contextual filtering. Based on the global analysis and the matched filtering blocks, different forms of order statistical filters were applied. These processing blocks yield an improved and adaptive fingerprint image processing method. Yang et al. [
16] proposed a novel and effective two-stage enhancement scheme in both the spatial domain and the frequency domain by learning from the underlying images. They first enhanced the fingerprint image in the spatial domain with a spatial ridge-compensation filter by learning from the images. With the help of the first step, the second stage filter, i.e., a frequency band-pass filter that was separable in the radial and angular frequency domains was employed. The experimental result showed that their algorithm is able to handle various input image contexts and achieves better results compared with some state-of-the-art algorithms over public databases and is able to improve the performances of fingerprint-authentication systems.
Fingerprints are closely related to genetics [
17]; however, in the fields of biostatics and psychology, there are currently no studies indicating any relationship between fingerprints and personality traits. Therefore, using fingerprints to induce personality traits is an undeveloped area in scientific research. If the corresponding relationship between fingerprints and personality traits could be determined, this would be an important contribution to science. Since personality traits have a certain degree of stability, continuity and uniqueness and the left/right hand fingerprints of each person are unique, the relationship between these two features is a worthwhile topic for in-depth research. The Big Five personalities have generated substantial interest among personality researchers [
3]. The Big Five is a model based on common language descriptors of personality. When factor analysis (a statistical technique) is applied to personality survey data, some words used to describe aspects of personality are often applied to the same person. These five factors are openness to experience (inventive/curious), conscientiousness (efficient/organized), extraversion (outgoing/energetic), agreeableness (friendly/compassionate) and neuroticism (sensitive/nervous).
The purpose in this study is to evaluate the generalizability of Big Five personality factor inventories as inducers of a common set of criteria, criteria representing classes of left and right thumb fingerprints. By assessing people using multiple criterion variables to measure the Big Five personality constructs, the same measure results normally will have the same personality constructs. If the Big Five inventories are all designed to measure equivalent dimensions of personality, then they should show a nontrivial amount of agreement in the variables they are able to induce. Constructing valid measures of personality variables should induce fingerprint classes, assuming those classes have personality determinants. This is especially true of Big Five inventories because those factors are presumed to account for most of the personality-based variation in fingerprints.
This study used classification technology to derive eight fingerprint types and combined these with questionnaire survey results to construct a new “System for Induction of Personality Traits from Fingerprints”. Following the research of Costa and McCare [
18], this study also summarized 14 personality constructs with Eigen values greater than one from the “Big Five” personality factors. We performed a principal components analysis of the data and found 14 components with Eigen values larger than one. Then, we created 14 scales each comprised of one of the 14 groups of items indicating the 14 components with Eigen values larger than one. The prototype of this system was modified and completed based on the fingerprints and questionnaire feedback of 362 test subjects. This study recruited a separate group of 351 subjects for the live testing of the system. The experimental results showed that the thumbprint types of the left and right hands were correlated with personality traits. Subjects in the left loop/right loop fingerprint category accounted for the largest group (41.8%). The second largest group was the S-type/S-type (twin loop/twin loop) type (13.5%), followed by the eddy/eddy type (12.1%). The personality traits of the latter two groups showed significant differences in some constructs.
Whilst better known in medication, double blind experiments are adopted in this paper. Surveys with questionnaires are used to keep credibility so the chance of observer’s bias can be minimized. The framework of the following sections in this paper is as follows:
Section 2: research framework and flow figure, expansion of the “Big Five” personality factors into 14 constructs, design of the personality traits questionnaire and the “System for Induction of Personality Traits from Fingerprints”;
Section 3: statistical analysis and post-test verification of the survey results on the relationship between personality traits and finger classification;
Section 4: conclusions.
3. Experimental Results
This study recovered 282 valid questionnaires. Initial results indicated that in left/right hand fingerprint types, the right hand fingerprints did not show arch type; arch type was also not found in some of the left hand fingerprints. Among the fingerprint types, the left loop/right loop type accounted for the largest group of test subjects (118 subjects), followed by the S-type/S-type (38 subjects), eddy/eddy type (34 subjects) and whorl/whorl type (28 subjects). The summary of type numbers is shown in
Table 2 and
Figure 3. In the questionnaire, 1–5 points each were assigned to the Likert scale options (strongly disagree, disagree, neither agree, nor disagree, agree, strongly agree). This study calculated the mean and standard deviation of each of the 14 personality constructs. Based on the responses on Likert items, this study derived the interrelationship between fingerprint type and personality construct; for details, please see
Table 3.
Table 3 shows that S-type/right Loop had the highest overall average in the 14 constructs, indicating that subjects with this fingerprint type demonstrated significant inclination in personality traits. The overall average of arch/whorl was the second highest in the 14 constructs, particularly with regard to the traits of “socially harmonious method of operation”, “concern for others’ well-being”, “enthusiastic attitude” and “strong sense of responsibility”; the overall average in terms of these four constructs even exceeded that of S-type/right loop. This indicated that individuals with this fingerprint type have outstanding leadership qualities. Additionally, to summarize the distribution trend of fingerprint type and personality constructs, a sample distributed clustering image of fingerprint type and personality traits is shown in
Figure 4, using four of the fingerprint types that accounted for a higher number of subjects and two personality traits. This is a sample distributed chart of personality traits based on fingerprint type, using four of the fingerprint types that accounted for a number of subjects and two personality traits (1–5 points each were assigned to the Likert scale options).