pith. sign in

arxiv: 1906.11960 · v1 · pith:ZMPJ6KRSnew · submitted 2019-06-27 · 💻 cs.HC · cs.CV· cs.LG

Studying the Impact of Mood on Identifying Smartphone Users

Khadija Zanna , Sayde King , Tempestt Neal , Shaun Canavan This is my paper

Pith reviewed 2026-05-25 14:24 UTC · model grok-4.3

classification 💻 cs.HC cs.CVcs.LG
keywords smartphone user identificationbehavioral biometricsmood impactintra-person variationStudentLife datasetmobile authenticationusage patterns
0
0 comments X

The pith

Mood does not negatively affect smartphone user identification performance

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper examines the effect of mood on smartphone user identification by testing performance with and without samples from periods when subjects felt happy, upset, or stressed. Using data from 19 subjects, it finds that removing these samples reduces identification accuracy. This contradicts the assumption in behavioral biometrics that mood is a harmful source of variation. The authors also observe that mood correlates with changes in certain usage patterns but conclude that biometric systems are not likely influenced negatively by mood.

Core claim

Performance worsens when removing samples generated when subjects may be happy, upset, or stressed. Thus, there is no indication that mood negatively impacts performance, although changes in smartphone usage patterns may correlate with mood, including changes in locking, audio, location, calling, homescreen, and e-mail habits.

What carries the argument

Comparison of identification performance on the full dataset versus the dataset with mood-labeled samples removed.

If this is right

  • Identification accuracy decreases without the mood-associated samples.
  • Mood correlates with variations in locking, audio, location, calling, homescreen, and e-mail habits.
  • Biometric systems may not need to treat mood as a negative source of variation.
  • Mood is a source of intra-person variation but one that does not reduce identification performance.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Training models on mood-diverse data could improve robustness if the performance benefit holds.
  • The same removal test could be applied to other candidate sources of behavioral variation such as time of day.
  • Systems might treat mood shifts as an additional usable signal rather than noise to be filtered.

Load-bearing premise

The mood states are reliably and accurately labeled for the samples in the dataset.

What would settle it

Repeating the removal experiment on a new dataset where mood is measured and labeled independently to check whether performance still declines without those samples.

Figures

Figures reproduced from arXiv: 1906.11960 by Khadija Zanna, Sayde King, Shaun Canavan, Tempestt Neal.

Figure 2
Figure 2. Figure 2: provides F-scores when using only the sam￾ples corresponding with mood. Here, we see more significant effects that may be associated with hap￾piness, upsetnees, and stress. First, none of these experiments yield results that outperform the use of all samples. We do note, however, that samples as￾sociated with mood inferred daily (D) perform worse than those inferred by the hour (H). Importantly, however, w… view at source ↗
Figure 1
Figure 1. Figure 1: F-scores for each experiment that excludes mood-related samples across different training set sizes. Thus far, we have evaluated identification perfor￾mance by removing samples associated with mood [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 3
Figure 3. Figure 3: Pearson’s correlation coefficients on EMA [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
read the original abstract

This paper explores the identification of smartphone users when certain samples collected while the subject felt happy, upset or stressed were absent or present. We employ data from 19 subjects using the StudentLife dataset, a dataset collected by researchers at Dartmouth College that was originally collected to correlate behaviors characterized by smartphone usage patterns with changes in stress and academic performance. Although many previous works on behavioral biometrics have implied that mood is a source of intra-person variation which may impact biometric performance, our results contradict this assumption. Our findings show that performance worsens when removing samples that were generated when subjects may be happy, upset, or stressed. Thus, there is no indication that mood negatively impacts performance. However, we do find that changes existing in smartphone usage patterns may correlate with mood, including changes in locking, audio, location, calling, homescreen, and e-mail habits. Thus, we show that while mood is a source of intra-person variation, it may be an inaccurate assumption that biometric systems (particularly, mobile biometrics) are likely influenced by mood.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper examines the effect of mood on smartphone user identification performance using the StudentLife dataset (19 subjects). It reports that removing samples labeled as collected during happy, upset, or stressed states degrades identification accuracy, from which it concludes there is no indication that mood negatively impacts performance. It additionally observes correlations between mood and changes in usage patterns (locking, audio, location, calling, homescreen, email).

Significance. If the central empirical claim holds after proper controls, the result would challenge the common assumption in behavioral biometrics that mood-induced intra-person variation harms identification accuracy and could simplify requirements for mobile biometric systems. The use of a named public dataset is a positive attribute.

major comments (2)
  1. [Results] Results section: the headline comparison (performance with vs. without mood-labeled samples) lacks a control ablation that removes an equal number of randomly chosen samples (or time-of-day/user-matched samples). This control is load-bearing for the claim that the observed degradation is attributable to mood rather than reduced data volume; without it the interpretation that 'mood does not negatively impact performance' cannot be supported.
  2. [Methodology] Methodology: the identification algorithm, feature set, classifier, cross-validation scheme, per-mood sample counts, and statistical tests are not described in sufficient detail to assess whether the reported performance differences are reliable or replicable.
minor comments (1)
  1. [Abstract] Abstract and introduction: the phrasing 'our findings show that performance worsens when removing samples...' should be qualified to note that the comparison is uncontrolled for sample size.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments. We address each major point below and will revise the manuscript accordingly to strengthen the empirical claims and improve replicability.

read point-by-point responses

  1. Referee: [Results] Results section: the headline comparison (performance with vs. without mood-labeled samples) lacks a control ablation that removes an equal number of randomly chosen samples (or time-of-day/user-matched samples). This control is load-bearing for the claim that the observed degradation is attributable to mood rather than reduced data volume; without it the interpretation that 'mood does not negatively impact performance' cannot be supported.

    Authors: We agree that the current comparison does not isolate the effect of mood from the effect of reduced sample volume. We will add a control ablation that removes an equal number of randomly chosen samples (and, where feasible, time-of-day or user-matched samples) and report the resulting identification accuracies alongside the mood-removal results. This will allow readers to assess whether the observed degradation is specifically attributable to mood-labeled samples. revision: yes

  2. Referee: [Methodology] Methodology: the identification algorithm, feature set, classifier, cross-validation scheme, per-mood sample counts, and statistical tests are not described in sufficient detail to assess whether the reported performance differences are reliable or replicable.

    Authors: We acknowledge the need for greater methodological transparency. In the revised manuscript we will expand the Methodology section to specify the exact identification algorithm and feature extraction pipeline, the classifier and its hyperparameters, the cross-validation scheme (including how train/test splits respect subject identity), the per-mood and per-subject sample counts, and the statistical tests used to evaluate performance differences. revision: yes

Circularity Check

0 steps flagged

No circularity: straightforward empirical ablation on public dataset

full rationale

The paper conducts a direct empirical comparison of smartphone user identification performance on the StudentLife dataset, with and without samples labeled as happy/upset/stressed. No equations, fitted parameters, derivations, or self-citations appear in the load-bearing steps; the central claim follows from counting and classifying existing data points rather than reducing to any input by construction. The analysis is self-contained against the external benchmark of the named public dataset.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No mathematical model or derivation is present; the work is an empirical comparison using an existing public dataset and standard behavioral biometric techniques.

pith-pipeline@v0.9.0 · 5718 in / 1078 out tokens · 35518 ms · 2026-05-25T14:24:31.040062+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages

  1. [1]

    Asselbergs, J

    J. Asselbergs, J. Ruwaard, M. Ejdys, N. Schrader, M. Sijbrandij, and H. Riper. Mobile phone-based un- obtrusive ecological momentary assessment of day- to-day mood: an explorative study. Journal of med- ical Internet research, 18(3), 2016

    work page 2016

  2. [2]

    Bardram, Mads Frost, K´ aroly Sz´ ant´ o, and Gabriela Marcu

    Jakob E. Bardram, Mads Frost, K´ aroly Sz´ ant´ o, and Gabriela Marcu. The monarca self-assessment sys- tem: A persuasive personal monitoring system for bipolar patients. In Proceedings of the 2Nd ACM SIGHIT International Health Informatics Sympo- sium, IHI ’12, pages 21–30, New York, NY, USA,

    work page

  3. [3]

    Biometric authentication on iphone and android: Usability, perceptions, and influences on adoption

    Rasekhar Bhagavatula, Blase Ur, Kevin Iacovino, Su Mon Kywe, Lorrie Faith Cranor, and Marios Savvides. Biometric authentication on iphone and android: Usability, perceptions, and influences on adoption. In In Proc. USEC, 2015

    work page 2015

  4. [4]

    Au- tomatic usability and stress analysis in mobile bio- metrics

    Ramon Blanco-Gonzalo, Raul Sanchez-Reillo, Os- car Miguel-Hurtado, and Eric Bella-Pulgarin. Au- tomatic usability and stress analysis in mobile bio- metrics. Image and Vision Computing , 32(12):1173 – 1180, 2014

    work page 2014

  5. [5]

    Wang, and Yi-Jong Jeng

    Mei-Yen Chen, Edward K. Wang, and Yi-Jong Jeng. Adequate sleep among adolescents is positively asso- ciated with health status and health-related behav- iors. BMC Public Health , 6(1):59, Mar 2006

    work page 2006

  6. [6]

    I feel like i’m taking selfies all day!: Towards understanding bio- metric authentication on smartphones

    Alexander De Luca, Alina Hang, Emanuel von Zezschwitz, and Heinrich Hussmann. I feel like i’m taking selfies all day!: Towards understanding bio- metric authentication on smartphones. In Proceed- ings of the 33rd Annual ACM Conference on Human Factors in Computing Systems, CHI ’15, pages 1411– 1414, New York, NY, USA, 2015. ACM

    work page 2015

  7. [7]

    Basic moods

    Craig DeLancey. Basic moods. Philosophical Psy- chology, 19(4):527–538, 2006

    work page 2006

  8. [8]

    Relationship of smartphone use severity with sleep quality, depression, and anxiety in uni- versity students

    Kadir Demirci, Mehmet Akgnl, and Abdullah Akpinar. Relationship of smartphone use severity with sleep quality, depression, and anxiety in uni- versity students. Journal of Behavioral Addictions , 4(2):85–92, 2015

    work page 2015

  9. [9]

    The effects of weather on daily mood: A multilevel approach

    Jaap JA Denissen, Ligaya Butalid, Lars Penke, and Marcel AG Van Aken. The effects of weather on daily mood: A multilevel approach. Emotion, 8(5):662, 2008

    work page 2008

  10. [10]

    Drira, B

    H. Drira, B. Ben Amor, A. Srivastava, M. Daoudi, and R. Slama. 3d face recognition under expres- sions, occlusions, and pose variations. IEEE Trans- actions on Pattern Analysis and Machine Intelli- gence, 35(9):2270–2283, Sep. 2013

    work page 2013

  11. [11]

    Affect, mood, and emotion

    Panteleimon Ekkekakis. Affect, mood, and emotion. Measurement in sport and exercise psychology , 321, 2012

    work page 2012

  12. [12]

    A density-based algorithm for discover- ing clusters a density-based algorithm for discover- ing clusters in large spatial databases with noise

    Martin Ester, Hans-Peter Kriegel, J¨ org Sander, and Xiaowei Xu. A density-based algorithm for discover- ing clusters a density-based algorithm for discover- ing clusters in large spatial databases with noise. In Proceedings of the Second International Conference on Knowledge Discovery and Data Mining , KDD’96, pages 226–231. AAAI Press, 1996. 7

    work page 1996

  13. [13]

    Huan Feng, Kassem Fawaz, and Kang G. Shin. Con- tinuous authentication for voice assistants. In Pro- ceedings of the 23rd Annual International Confer- ence on Mobile Computing and Networking , Mobi- Com ’17, pages 343–355, New York, NY, USA, 2017. ACM

    work page 2017

  14. [14]

    Joseph P. Forgas. Mood effects on decision mak- ing strategies. Australian Journal of Psychology , 41(2):197–214, 1989

    work page 1989

  15. [15]

    Frank, R

    M. Frank, R. Biedert, E. Ma, I. Martinovic, and D. Song. Touchalytics: On the applicability of touch- screen input as a behavioral biometric for continuous authentication. IEEE Transactions on Information Forensics and Security, 8(1):136–148, Jan 2013

    work page 2013

  16. [16]

    Sleep and the affective response to stress and pain

    Nancy A Hamilton, Delwyn Catley, and Cynthia Karlson. Sleep and the affective response to stress and pain. Health Psychology, 26(3):288, 2007

    work page 2007

  17. [17]

    It’s a hard lock life: A field study of smart- phone (un)locking behavior and risk perception

    Marian Harbach, Emanuel von Zezschwitz, An- dreas Fichtner, Alexander De Luca, and Matthew Smith. It’s a hard lock life: A field study of smart- phone (un)locking behavior and risk perception. In 10th Symposium On Usable Privacy and Security (SOUPS 2014) , pages 213–230, Menlo Park, CA,

    work page 2014

  18. [18]

    A. K. Jain, A. Ross, and S. Prabhakar. An introduc- tion to biometric recognition. IEEE Transactions on Circuits and Systems for Video Technology, 14(1):4– 20, Jan 2004

    work page 2004

  19. [19]

    Jain, Arun A

    Anil K. Jain, Arun A. Ross, and Karthik Nandaku- mar. Introduction, pages 1–49. Springer US, Boston, MA, 2011

    work page 2011

  20. [20]

    Predicting students’ happiness from physi- ology, phone, mobility, and behavioral data

    Natasha Jaques, Sara Taylor, Asaph Azaria, Asma Ghandeharioun, Akane Sano, and Rosalind Pi- card. Predicting students’ happiness from physi- ology, phone, mobility, and behavioral data. In 2015 International Conference on Affective Comput- ing and Intelligent Interaction (ACII) , pages 222–

    work page 2015

  21. [21]

    Investigating user comprehension and risk perception of apple’s touch id technology

    Yousra Javed, Mohamed Shehab, and Emmanuel Bello-Ogunu. Investigating user comprehension and risk perception of apple’s touch id technology. In Proceedings of the 12th International Conference on Availability, Reliability and Security , ARES ’17, pages 35:1–35:6, New York, NY, USA, 2017. ACM

    work page 2017

  22. [22]

    Lane, and Lin Zhong

    Robert LiKamWa, Yunxin Liu, Nicholas D. Lane, and Lin Zhong. Moodscope: Building a mood sensor from smartphone usage patterns. In Proceeding of the 11th Annual International Conference on Mobile Systems, Applications, and Services , MobiSys ’13, pages 389–402, New York, NY, USA, 2013. ACM

    work page 2013

  23. [23]

    Y. Ma, B. Xu, Y. Bai, G. Sun, and R. Zhu. Daily mood assessment based on mobile phone sensing. In International conference on wearable and im- plantable body sensor networks, pages 142–147, 2012

    work page 2012

  24. [24]

    Wong, Sheng Wen, and Yang Xiang

    Weizhi Meng, Yu Wang, Duncan S. Wong, Sheng Wen, and Yang Xiang. Touchwb: Touch behavioral user authentication based on web browsing on smart- phones. Journal of Network and Computer Applica- tions, 117:1 – 9, 2018

    work page 2018

  25. [25]

    William N. Morris. Mood: The Frame of Mind . Springer, New York, NY, 1989

    work page 1989

  26. [26]

    Palfai and Peter Salovey

    Tibor P. Palfai and Peter Salovey. The influence of depressed and elated mood on deductive and induc- tive reasoning. Imagination, Cognition and Person- ality, 13(1):57–71, 1993

    work page 1993

  27. [27]

    V. M. Patel, R. Chellappa, D. Chandra, and B. Bar- bello. Continuous user authentication on mobile devices: Recent progress and remaining challenges. IEEE Signal Processing Magazine, 33(4):49–61, July 2016

    work page 2016

  28. [28]

    Detecting change points in the large-scale structure of evolving net- works

    Leto Peel and Aaron Clauset. Detecting change points in the large-scale structure of evolving net- works. In Twenty-Ninth AAAI Conference on Arti- ficial Intelligence, 2015

    work page 2015

  29. [29]

    Pratap, D

    A. Pratap, D. Atkins, B. Renn, M. Tanana, S. Mooney, and P. Anguera, J .and Are´ an. The ac- curacy of passive phone sensors in predicting daily mood. Depression and anxiety , 36(1):72–81, 2019

    work page 2019

  30. [30]

    A circumplex model of affect

    James A Russell. A circumplex model of affect. Jour- nal of personality and social psychology , 39(6):1161, 1980

    work page 1980

  31. [31]

    Identifying objective physiological markers and modifiable behaviors for self-reported stress and mental health status using wearable sensors and mobile phones: observational study

    Akane Sano, Sara Taylor, Andrew W McHill, An- drew JK Phillips, Laura K Barger, Elizabeth Kler- man, and Rosalind Picard. Identifying objective physiological markers and modifiable behaviors for self-reported stress and mental health status using wearable sensors and mobile phones: observational study. Journal of medical Internet research , 20(6), 2018

    work page 2018

  32. [32]

    Stone, and Michael R

    Saul Shiffman, Arthur A. Stone, and Michael R. Huf- ford. Ecological momentary assessment. Annual Re- view of Clinical Psychology , 4(1):1–32, 2008. PMID: 18509902

    work page 2008

  33. [33]

    Stone, Joshua M

    Arthur A. Stone, Joshua M. Smyth, Thomas Pick- ering, and Joseph Schwartz. Daily mood variability: Form of diurnal patterns and determinants of diur- nal patterns. Journal of Applied Social Psychology , 26(14):1286–1305, 1996

    work page 1996

  34. [34]

    Activity-aware mental stress detection using physiological sensors

    Feng-Tso Sun, Cynthia Kuo, Heng-Tze Cheng, Senaka Buthpitiya, Patricia Collins, and Martin Griss. Activity-aware mental stress detection using physiological sensors. In International conference on Mobile computing, applications, and services , pages 282–301. Springer, 2010

    work page 2010

  35. [35]

    Realizing the potential of mobile mental health: New methods for new data in psychiatry

    John Torous, Patrick Staples, and Jukka-Pekka On- nela. Realizing the potential of mobile mental health: New methods for new data in psychiatry. Current Psychiatry Reports, 17(8):61, Jun 2015

    work page 2015

  36. [36]

    Mobile sensing and support for people with depression: A pilot trial 8 in the wild

    Fabian Wahle, Tobias Kowatsch, Elgar Fleisch, Michael Rufer, and Steffi Weidt. Mobile sensing and support for people with depression: A pilot trial 8 in the wild. JMIR Mhealth Uhealth , 4(3):e111, Sep 2016

    work page 2016

  37. [37]

    Studentlife: assessing mental health, academic performance and behavioral trends of college students using smart- phones

    Rui Wang, Fanglin Chen, Zhenyu Chen, Tianxing Li, Gabriella Harari, Stefanie Tignor, Xia Zhou, Dror Ben-Zeev, and Andrew T Campbell. Studentlife: assessing mental health, academic performance and behavioral trends of college students using smart- phones. In Proceedings of the 2014 ACM interna- tional joint conference on pervasive and ubiquitous computing,...

    work page 2014

  38. [38]

    Hui Xu, Yangfan Zhou, and Michael R. Lyu. To- wards continuous and passive authentication via touch biometrics: An experimental study on smart- phones. In 10th Symposium On Usable Privacy and Security (SOUPS 2014), pages 187–198, Menlo Park, CA, 2014. USENIX Association. 9

    work page 2014