Przybylinski, E. (2026). Whatever you say: Changing transference-based problem behavior with if–then plans. Self and Identity. Advance online publication. https://doi.org/10.1080/15298868.2026.2613846


Przybylinski (2026) reports two experiments examining whether implementation intentions can prevent problematic behaviors triggered by transference. The theoretical logic is straightforward: subtle resemblance to a significant other is assumed to activate that person’s representation automatically, which can then influence memory, goals, and behavior. An if–then plan is proposed to prevent the activated representation from guiding subsequent behavior.

The principal concern is the credibility of the evidence on which this argument rests.

The article treats automatic behavioral priming as a well-established foundation. For example, it cites Bargh et al. (1996), Bargh et al. (2001), Chartrand and Bargh (1999), and related studies as evidence that contextual cues can automatically trigger overt behavior without awareness or intention. Yet behavioral priming is precisely one of the areas most affected by the replication crisis. The article does not discuss this change in evidential status. Thus, evidence that was considered persuasive when these studies were conducted is largely presented in 2026 as though subsequent replication failures had not occurred.

This matters because the two experiments are themselves products of that earlier research era. The author explicitly states that they were conducted as dissertation research in the late 2000s, before preregistration became standard. Study 1 included only 60 participants, or 20 participants in each of the three strategy conditions, despite testing interaction hypotheses. Study 2 included 47 participants. The manuscript provides a power justification, but because the studies were not preregistered, it is unclear whether the reported power analysis reflects a prospectively specified design decision. The reported “post hoc power” provides little additional information.

The results are remarkably successful. The focal interactions involving behavioral readiness, memory, and overt submissive behavior are all statistically significant in the predicted direction. The reported effects are also very large, frequently exceeding d = 1 and reaching d = 1.69 in Study 1. Across the major focal tests, the success rate is effectively 100%, while average observed power based on the reported effects is roughly 80%.

A perfect success rate is not impossible when power is 80%, but it is more successful than expected. Schimmack (2012) emphasized that unusually high success rates relative to estimated power can indicate that published effect sizes and success rates should not be taken at face value. Here, the outcomes within each experiment are dependent, so a simple excess-success or incredibility calculation would not be appropriate. With only two studies, there is no statistical smoking gun. Nevertheless, the combination of small samples, large effects, multiple significant focal outcomes, and absence of preregistration warrants substantial caution.

The historical timing makes this concern more important. These experiments were conducted approximately 15 years before their publication. During that interval, psychology experienced a replication crisis that directly challenged the credibility of the behavioral-priming literature on which the article relies. Yet the 2026 article does not supplement the old experiments with a contemporary, adequately powered, preregistered replication.

This is particularly striking because the central experiment is readily replicable. The author remains at the same institution where the original research was conducted, and the paradigm requires undergraduate participants rather than an unusually difficult population. A preregistered replication with a substantially larger sample could have provided highly informative evidence about whether the large effects observed in the original dissertation studies survive contemporary scrutiny.

The absence of such a replication changes how the evidence should be interpreted. The results are not invalid merely because they were collected before the replication crisis. But neither the reported effect sizes nor the perfect pattern of statistical success should be treated as reliable estimates of the underlying effects without independent replication.