Reduced cooperation in borderline personality disorder largely reflects an absence of consistent cooperators: a reanalysis of Saunders et al. (2015)
Abstract
Background. Interpersonal difficulties are a core feature of borderline personality disorder (BPD). Saunders et al. (2015) reported that, in an iterated Prisoner’s Dilemma against a Tit-for-Tat partner, women with BPD cooperated less than both women with euthymic bipolar disorder and healthy controls, particularly after rounds with mutual cooperation. However, a Tit-for-Tat partner simply copies the participant's previous choice, so the game states a participant encounters are determined by her own behaviour: someone who always cooperates experiences only mutual cooperation, whereas someone who frequently defects rarely does. Cooperation rates conditional on the previous outcome are therefore potentially confounded with overall cooperativeness, and a group difference after mutual cooperation need not reflect a different response to that outcome.
Methods. To test this possibility, we reanalysed the original data (n = 58; 19 with BPD, 39 controls [19 bipolar, 20 healthy]; all female). For each participant, we simulated the conditional cooperation rates expected if she cooperated with a fixed probability equal to her overall cooperation rate, irrespective of the previous outcome, given the transition structure imposed by Tit-for-Tat. We then compared observed with expected conditional rates in each game state.
Results. Once overall cooperation rates were accounted for, conditional cooperation did not differ between groups: rather, observed rates matched expected rates in every state except after the exploitation of a cooperating partner, where all groups defected more than expected (in BPD only nominally significant, and not after false discovery rate correction). Further, the observed lower overall cooperation in BPD compared to both control groups was driven by consistent cooperators: a third of controls (13 of 39) but no participant with BPD (0 of 19) cooperated on at least 90% of rounds. Excluding these consistent cooperators reduced the group difference by about three-quarters.
Conclusions. Women with BPD cooperated less overall, but this reflected an absence of consistently cooperating individuals rather than a specific failure to sustain mutual cooperation. More generally, analyses of iterated games against deterministic strategies must account for the state transitions induced by participants' own choices; otherwise, differences in overall cooperation can masquerade as differences in strategy.
Related
- Canine-Assisted Interventions and Symptoms Associated with Borderline Personality Disorder: A Scoping Review
- Piloting New Medications for Borderline Personality Disorder: The Use of Artificial Intelligence in Psychopharmacological Research
- Attachment, Emerging Borderline Personality Features and Mentalizing Capacity: A Comparative Analysis in Maltreated and Community Children
- Machine Learning in the Differential Diagnosis of Bipolar and Borderline Personality Disorders: A Scoping Review
- Borderline personality disorder, a diagnosis that cannot be challenged: If rejecting the diagnosis is a symptom, can the diagnosis ever be wrong? A response to Chazhoor and Younger (2026)
BibTeX
@misc{kupersmith2026reduced,
title = {{Reduced cooperation in borderline personality disorder largely reflects an absence of consistent cooperators: a reanalysis of Saunders et al. (2015)}},
author = {Benjamin James Kuper-Smith and Kate Saunders and Robert D Rogers and Christian Ruff},
howpublished = {PsyArXiv},
year = {2026},
month = sep,
doi = {10.31234/osf.io/wvup5_v1},
url = {https://osf.io/preprints/psyarxiv/wvup5_v1/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Reduced cooperation in borderline personality disorder largely reflects an absence of consistent cooperators: a reanalysis of Saunders et al. (2015)
AU - Benjamin James Kuper-Smith
AU - Kate Saunders
AU - Robert D Rogers
AU - Christian Ruff
PY - 2026
DA - 2026/09/30
PB - PsyArXiv
DO - 10.31234/osf.io/wvup5_v1
UR - https://osf.io/preprints/psyarxiv/wvup5_v1/
AB - Background. Interpersonal difficulties are a core feature of borderline personality disorder (BPD). Saunders et al. (2015) reported that, in an iterated Prisoner’s Dilemma against a Tit-for-Tat partner, women with BPD cooperated less than both women with euthymic bipolar disorder and healthy controls, particularly after rounds with mutual cooperation. However, a Tit-for-Tat partner simply copies the participant's previous choice, so the game states a participant encounters are determined by her own behaviour: someone who always cooperates experiences only mutual cooperation, whereas someone who frequently defects rarely does. Cooperation rates conditional on the previous outcome are therefore potentially confounded with overall cooperativeness, and a group difference after mutual cooperation need not reflect a different response to that outcome. Methods. To test this possibility, we reanalysed the original data (n = 58; 19 with BPD, 39 controls [19 bipolar, 20 healthy]; all female). For each participant, we simulated the conditional cooperation rates expected if she cooperated with a fixed probability equal to her overall cooperation rate, irrespective of the previous outcome, given the transition structure imposed by Tit-for-Tat. We then compared observed with expected conditional rates in each game state. Results. Once overall cooperation rates were accounted for, conditional cooperation did not differ between groups: rather, observed rates matched expected rates in every state except after the exploitation of a cooperating partner, where all groups defected more than expected (in BPD only nominally significant, and not after false discovery rate correction). Further, the observed lower overall cooperation in BPD compared to both control groups was driven by consistent cooperators: a third of controls (13 of 39) but no participant with BPD (0 of 19) cooperated on at least 90% of rounds. Excluding these consistent cooperators reduced the group difference by about three-quarters. Conclusions. Women with BPD cooperated less overall, but this reflected an absence of consistently cooperating individuals rather than a specific failure to sustain mutual cooperation. More generally, analyses of iterated games against deterministic strategies must account for the state transitions induced by participants' own choices; otherwise, differences in overall cooperation can masquerade as differences in strategy.
ER -