Borderline Personality Disorder
The Latest Theory of Borderline Personality Disorder
New research uses logic to understand people with borderline disorder.
Posted April 27, 2024 Reviewed by Lybi Ma
Key points
- Theories of borderline personality disorder focus on early emotional experiences as the cause of splitting.
- A new approach uses statistical modeling to see what factors influence the tendency toward splitting.
- People with this disorder can use their experiences to help overcome splitting and achieve integration.
When you think of someone with borderline personality disorder, it’s likely that you imagine their tendency to let their emotions veer out of control, become overly intrusive, and alternate between loving and hating the same individual, or what’s known as “splitting.”
Perhaps you have an in-law who you know has this diagnosis, and as much as you would like to show understanding and empathy, it’s difficult when they decide to make you the target of the hate side of the splitting equation. If only you could figure out what’s behind these stormy changes of heart, perhaps you could get along better, if not become a trusted source of help. New research uses a logical analysis to provide some clues.
The Logic of Splitting in Borderline Personality Disorder
According to the University of College London’s Giles Story and colleagues (2024), traditional theories of this disorder regard splitting as an offshoot of unhealthy early emotional development, in which “an infant structures its experience by discriminating between positive and negative affect… [leading to] disconnected states of extreme satisfaction and frustration.” (p. 750) In ordinary development, individuals learn to bring these two states into balance as the bonds with their caregivers become stable and enduring, regardless of whether their needs are satisfied or frustrated. In BPD, though, this integration fails to occur, and individuals come to acquire split representations not just of other people, but of themselves.
Tracing the development of these “object relations” approaches to BPD, Story and colleagues maintain that these approaches miss the mark by considering only emotional aspects of splitting and not potential cognitive ones. Instead, the UCL authors propose a framework of probabilistic inference. Perhaps people with BPD form the wrong conclusions when they try to establish the causes of other people’s behaviors.
Everyone engages in this type of speculation, whether BPD or not. Even a simple act such as someone smiling at you could have you wonder what’s behind their apparent good intentions. Did you do something they liked, which would be a “situational” cause, or are they just people who tend to smile (“dispositional” cause)? You might then go on to observe their future behavior which would allow you to settle the question.
There are many possible variants of the situational versus dispositional attributional process. If you like someone, you’ll be more inclined to view good behavior as dispositional. If someone generally behaves a certain way and then suddenly shows the opposite behavior, it must be situational. For the most part, these reasoning processes are rational, allowing individuals to test “hypotheses” about why others behave as they do. In BPD, all-or-nothing reasoning sets in, in which people tend to view the causes of other people’s behavior in overly simplistic, all-or-nothing ways.
A Statistical Modeling Approach to Splitting
In ordinary experience, Story and colleagues propose, people go through statistical thinking known as “Bayesian,” in which they change the probabilities of a person’s behavior reflecting dispositional factors based on behaviors as they unfold over time. The first time you see a stranger smile, you have no idea whether they are nice, or not. However, as you see them continuing to smile regardless of what’s happening around them, the odds increasingly shift in favor of a dispositional attribution. If people with BPD can’t make that calculation, they’ll judge someone’s personality based on a snap judgment that won’t change over time. Splitting, in this view, becomes “a distorted causal inference.” They don’t, as the authors propose, update a “person prior” (p. 757). Once nice, always nice, and vice versa. The “split priors,” in which people are all bad or all good become “impervious to learning." (p. 762)
The UCL authors ran statistical models on data from a prior study of participants with BPD and non-BPD controls given the simulated task of judging the moral character of “Person A” toward “Person B.” Ordinarily if you were making such a judgment with no “prior person,” you would have to wait to see what Person A actually did in a given situation and then update your judgments. By running a model against the actual data from participants as they judged whether Person A was a good or a bad “agent,” Story and colleagues could determine whether the groups differed in their ability to draw Bayesian inferences.
Not only did the findings confirm the predictions of differences in attributional processes between BPD and non-BPD groups, but the authors were also able to run a model to show how the tendency to engage in splitting could potentially be modified. Part of the problem in making attributions in general is that sometimes the external context in which people act isn’t all that clear-cut. The lack of clarity only feeds into the tendency of people with BPD to draw conclusions based on their initial impressions of others. However, by creating a model in which precise information is available about Person A’s behavior, attributions based on splitting diminish and a more integrated, and realistic, set of judgments can emerge.
Turning Statistical Modeling into Intervention
The job of intervention is to help “split priors” to be updated by data from experience, using the processes described through the Bayesian modeling. Consider the example provided by the authors. A person with BPD might jump to the conclusion that if someone they’re supposed to meet is late, this proves that “they hate me.” Intervention can help the individual take extra external information into account (such as heavy traffic). Training the individual to be more sensitive to situational factors would reduce the tendency to make judgments based not on what people do, but on how they imagine people to be (all good or all bad).
In general, it is a good idea to let the evidence influence your judgments of people, but for individuals who tend to block out the data from experience, this type of intervention could help the person with BPD become more attuned to what the people around them are actually doing. Additionally, based on the Bayesian model, in which your judgments about probability should shift depending on prior outcomes, training people to monitor their conclusions about people over time would help individuals integrate positive and negative judgments rather than shift toward one or the other.
To sum up, whatever early experiences shape the individual with BPD are less important, in this approach, than the logic of the decisions the individual makes in the here and now. Giving people with this disorder the power and knowledge to observe themselves over time can help build the fulfillment that comes from an integrated view of the people in their lives.
Facebook image: J Walters/Shutterstock
References
Story, G. W., Smith, R., Moutoussis, M., Berwian, I. M., Nolte, T., Bilek, E., Siegel, J. Z., & Dolan, R. J. (2024). A social inference model of idealization and devaluation. Psychological Review, 131(3), 749-780. https://doi.org/10.1037/rev0000430