Explainable AI for Psychological Profiling from Behavioral Data: An Application to Big Five Personality Predictions from Financial Transaction Records

Every step we take in the digital world leaves behind a record of our behavior; a digital footprint. Research has suggested that algorithms can translate these digital footprints into accurate estimates of psychological characteristics, including personality traits, mental health or intelligence. The mechanisms by which AI generates these insights, however, often remain opaque. In this paper, we show how Explainable AI (XAI) can help domain experts and data subjects validate, question, and improve models that classify psychological traits from digital footprints. We elaborate on two popular XAI methods (rule extraction and counterfactual explanations) in the context of Big Five personality predictions (traits and facets) from financial transactions data (N = 6,408). First, we demonstrate how global rule extraction sheds light on the spending patterns identified by the model as most predictive for personality, and discuss how these rules can be used to explain, validate, and improve the model. Second, we implement local rule extraction to show that individuals are assigned to personality classes because of their unique financial behavior, and there exists a positive link between the model’s prediction confidence and the number of features that contributed to the prediction. Our experiments highlight the importance of both global and local XAI methods. By better understanding how predictive models work in general as well as how they derive an outcome for a particular person, XAI promotes accountability in a world in which AI impacts the lives of billions of people around the world.

Download Full-text

Disentangling principled and opportunistic motives for reacting to injustice: A genetically-informed exploration of justice sensitivity

10.1101/2020.06.10.143925 ◽

2020 ◽

Author(s):

Nikolai Haahjem Eftedal ◽

Thomas Haarklau Kleppestø ◽

Nikolai Olavi Czajkowski ◽

Jennifer Sheehy-Skeffington ◽

Espen Røysamb ◽

...

Keyword(s):

Big Five ◽

Social Dominance ◽

Social Dominance Orientation ◽

Moral Judgments ◽

Big Five Personality ◽

Psychological Traits ◽

Twin Registry ◽

Justice Sensitivity ◽

Increased Sensitivity ◽

Single Set

AbstractMoral judgments may be driven by both principled and opportunistic motivations. Being morally principled is to consistently adhere to a single set of rules about morality and justice. Opportunistic morality rather involves selectively enforcing rules when they are beneficial to one’s interests. These two kinds of motivations sometimes pull in the same direction, other times not. Prior studies on moral motivations have mostly focused on principled morality. Opportunistic morality, along with its phenotypic and genetic correlates, remains largely unexamined. Here, utilizing a sample from the Norwegian Twin Registry, consisting of 312 monozygotic-and 298 dizygotic twin pairs (N = 1220), we measure people’s propensity to react to injustice as victims, observers, beneficiaries, and perpetrators of injustice, using the Justice Sensitivity scale. Our genetically informative sample allows a biometric modeling approach that provides increased stringency in inferring latent psychological traits. We find evidence for two substantially heritable traits explaining correlations between Justice Sensitivity facets, which we interpret as a principled justice sensitivity (h2 = .45) leading to increased sensitivity to injustices of all categories, and an opportunistic justice sensitivity (h2 = .69) associated with increased victim sensitivity and a decreased propensity to feel guilt from being a perpetrator. These heritable justice traits share a genetic substrate with broad strategies for cooperation (as measured by altruism and trust) and for selectively benefitting oneself over the adaptive interests of others (as measured by social dominance orientation and support for monopolizing territory and resources), and differ genetically and phenotypically from Big Five personality traits.

Download Full-text

A Study on the Impact of Big Five Personality Traits on Consciousness

International Journal of Indian Psychology ◽

10.25215/0302.180 ◽

2016 ◽

Vol 3 (2) ◽

Author(s):

Akriti Srivastava ◽

Anamika Mishra

Keyword(s):

Personality Traits ◽

Big Five ◽

Big Five Personality Traits ◽

Big Five Personality ◽

Psychological Traits ◽

Significant Difference ◽

Males And Females ◽

Neo Five Factor Inventory ◽

The Individual ◽

The Impact

Personality is the set of psychological traits and mechanisms within the individual that are organized and relatively enduring and that influence their interactions with, and adaptations to, the intra psychic, physical, and social environments. Personality is the factor which influences other functioning of the individuals, keeping this view in mind its relation with consciousness was thought to be explored, which is awareness, the ability to experience or to feel, wakefulness, having a sense of selfhood, and the executive control system of the mind. Hence, this study was designed to understand the relationship between big five personality traits and consciousness, for this, the data was collected from 200 working and non-working males and females in the age range of 22 to 30 years. Sample was chosen from Banasthali Vidyapith, Tonk; Rajiv Gandhi Technical Institute, Raibarely; and Wipro Company, Bangalore. Measures used were self-consciousness scale (Fenigstein, Scheier &Buss 1975) which is a 23 item questionnaire and personality questionnaire: NEO Five-Factor Inventory-3 (NEO-FFI-3). This questionnaire was developed by Paul T. Costa, Jr and Robert R. McCrae (1989). The result indicated that openness and conscientiousness are significantly positively related with consciousness and openness emerged out as the significant predictor of consciousness. Also the significant difference was found between males and females in terms of consciousness. This study is unique in its endeavor and creates scope for further exploration in this field.

Download Full-text

Further Validity of the Japanese Version of the Ten Item Personality Inventory (TIPI-J)

Journal of Individual Differences ◽

10.1027/1614-0001/a000145 ◽

2014 ◽

Vol 35 (4) ◽

pp. 236-244 ◽

Cited By ~ 16

Author(s):

Atsushi Oshio ◽

Shingo Abe ◽

Pino Cutrone ◽

Samuel D. Gosling

Keyword(s):

Big Five ◽

Content Validity ◽

Structural Equation ◽

English Language ◽

Full Range ◽

Reliability And Validity ◽

Japanese Version ◽

Personality Inventory ◽

Equation Modeling ◽

Big Five Personality

The Ten Item Personality Inventory (TIPI; Gosling, Rentfrow, & Swann, 2003 ) is a widely used very brief measure of the Big Five personality dimensions. Oshio, Abe, and Cutrone (2012) have developed a Japanese version of the TIPI (TIPI-J), which demonstrated acceptable levels of reliability and validity. Until now, all studies examining the validity of the TIPI-J have been conducted in the Japanese language; this reliance on a single language raises concerns about the instrument’s content validity because the instrument could demonstrate reliability (e.g., retest) and some forms of validity (e.g., convergent) but still not capture the full range of the dimensions as originally conceptualized in English. Therefore, to test the content validity of the Japanese TIPI with respect to the original Big Five formulation, we examine the convergence between scores on the TIPI-J and scores on the English-language Big Five Inventory (i.e., the BFI-E), an instrument specifically designed to optimize Big Five content coverage. Two-hundred and twenty-eight Japanese undergraduate students, who were all learning English, completed the two instruments. The results of correlation analyses and structural equation modeling demonstrate the theorized congruence between the TIPI-J and the BFI-E, supporting the content validity of the TIPI-J.

Download Full-text

Intelligence and Personality

Journal of Individual Differences ◽

10.1027/1614-0001/a000311 ◽

2020 ◽

Vol 41 (3) ◽

pp. 124-132

Author(s):

Marc-André Bédard ◽

Yann Le Corff

Keyword(s):

Personality Traits ◽

Big Five ◽

Real Life ◽

Big Five Personality Traits ◽

Original Study ◽

Life Assessment ◽

Big Five Personality ◽

Verbal Intelligence ◽

Nonverbal Intelligence ◽

French Canadian

Abstract. This replication and extension of DeYoung, Quilty, Peterson, and Gray’s (2014) study aimed to assess the unique variance of each of the 10 aspects of the Big Five personality traits ( DeYoung, Quilty, & Peterson, 2007 ) associated with intelligence and its dimensions. Personality aspects and intelligence were assessed in a sample of French-Canadian adults from real-life assessment settings ( n = 213). Results showed that the Intellect aspect was independently associated with g, verbal, and nonverbal intelligence while its counterpart Openness was independently related to verbal intelligence only, thus replicating the results of the original study. Independent associations were also found between Withdrawal, Industriousness and Assertiveness aspects and verbal intelligence, as well as between Withdrawal and Politeness aspects and nonverbal intelligence. Possible explanations for these associations are discussed.

Download Full-text

Using Principal Component Scores to Enhance the Validity and Reliability of Big Five Personality Measures

Journal of Individual Differences ◽

10.1027/1614-0001/a000225 ◽

2017 ◽

Vol 38 (2) ◽

pp. 83-93

Author(s):

Jeffrey M. Cucina ◽

Nicholas L. Vasilopoulos ◽

Arwen H. DeCostanza

Keyword(s):

Big Five ◽

Discriminant Validity ◽

Reliability And Validity ◽

Principal Component ◽

Big Five Personality ◽

Validity And Reliability ◽

Scoring Method ◽

Five Factors ◽

Big Five Factors ◽

Component Scores

Abstract. Varimax rotated principal component scores (VRPCS) have previously been offered as a possible solution to the non-orthogonality of scores for the Big Five factors. However, few researchers have examined the reliability and validity of VRPCS. To address this gap, we use a lab study and a field study to investigate whether using VRPCS increase orthogonality, reliability, and criterion-related validity. Compared to the traditional unit-weighting scoring method, the use of VRPCS enhanced the reliability and discriminant validity of the Big Five factors, although there was little improvement in criterion-related validity. Results are discussed in terms of the benefit of using VRPCS instead of traditional unit-weighted sum scores.

Download Full-text

How Do People Share Their Positive Events?

Journal of Individual Differences ◽

10.1027/1614-0001/a000212 ◽

2016 ◽

Vol 37 (4) ◽

pp. 250-259 ◽

Cited By ~ 10

Author(s):

Cara A. Palmer ◽

Meagan A. Ramsey ◽

Jennifer N. Morey ◽

Amy L. Gentzler

Keyword(s):

Cluster Analysis ◽

Communication Technology ◽

Big Five ◽

Well Being ◽

Big Five Personality Traits ◽

Big Five Personality ◽

Positive Events ◽

Growing Body ◽

Sharing Patterns ◽

Practical Implications

Abstract. Research suggests that sharing positive events with others is beneficial for well-being, yet little is known about how positive events are shared with others and who is most likely to share their positive events. The current study expanded on previous research by investigating how positive events are shared and individual differences in how people share these events. Participants (N = 251) reported on their likelihood to share positive events in three ways: capitalizing (sharing with close others), bragging (sharing with someone who may become jealous or upset), and mass-sharing (sharing with many people at once using communication technology) across a range of positive scenarios. Using cluster analysis, five meaningful profiles of sharing patterns emerged. These profiles were associated with gender, Big Five personality traits, narcissism, and empathy. Individuals who tended to brag when they shared their positive events were more likely to be men, reported less agreeableness, less conscientiousness, and less empathy, whereas those who tended to brag and mass-share reported the highest levels of narcissism. These results have important theoretical and practical implications for the growing body of research on sharing positive events.

Download Full-text

Personality of Clown Doctors

Journal of Individual Differences ◽

10.1027/1614-0001/a000187 ◽

2016 ◽

Vol 37 (1) ◽

pp. 49-55 ◽

Cited By ~ 5

Author(s):

Alberto Dionigi

Keyword(s):

General Population ◽

Personality Traits ◽

Big Five ◽

Normal Population ◽

Big Five Personality Traits ◽

Published Data ◽

Big Five Personality

Abstract. In recent years, both professional and volunteer clowns have become familiar in health settings. The clown represents a peculiar humorist’s character, strictly associated with the performer’s own personality. In this study, the Big Five personality traits (BFI) of 155 Italian clown doctors (130 volunteers and 25 professionals) were compared to published data for the normal population. This study highlighted specific differences between clown doctors and the general population: Clown doctors showed higher agreeableness, conscientiousness, openness, and extraversion, as well as lower neuroticism compared to other people. Moreover, specific differences emerged comparing volunteers and professionals: Professional clowns showed significantly lower in agreeableness compared to their unpaid colleagues. The results are also discussed with reference to previous studies conducted on groups of humorists. Clowns’ personalities showed some peculiarities that can help to explain the facility for their performances in the health setting and that are different than those of other groups of humorists.

Download Full-text