Addiction is linked to inconsistent decision-making, not ignoring consequences

People who regularly use drugs may not simply ignore negative consequences—but may instead struggle to consistently act on them, according to new research published in Translational Psychiatry.

Traditionally, scientists have argued that individuals with more severe substance use become less sensitive to negative consequences such as health problems or financial loss. However, real-world decisions are rarely that simple. People often face competing costs—for example, the discomfort of withdrawal or loneliness if they stop using versus the long-term harm of continuing. These decisions also take place in environments that can be stable or constantly changing, adding another layer of complexity.

Hence, the researchers behind this study wanted to move beyond simplified models of decision-making. Instead of focusing only on rewards or single types of cost, they examined how people compare multiple costs and how their decisions shift depending on the stability of their environment.

Led by Sonia G. Ruiz, the Yale University team recruited 137 adults aged 18 to 65 from the community, 75% of whom had a history of regular substance use. To measure substance use severity, the researchers calculated each participant’s cumulative lifetime “years of regular use” (defined as using a substance three or more times a week).

Participants then completed a computer-based task designed to mimic real-world decision-making under uncertainty. In each of 200 rounds, they chose between two options—represented as cards—that could result in losing random, small amounts of money ranging from one to five dollars. The goal was to minimize losses by learning which option was safer.

The task included two different conditions. In the “stable” condition, the likelihood of losing money stayed the same over time, meaning participants could rely on consistent patterns. In the “volatile” condition, the probabilities frequently changed, requiring participants to adapt quickly and pay more attention to recent outcomes. Participants were not told when these changes occurred, forcing them to learn through experience.

As expected, most participants gradually improved their choices by learning from feedback. They were more likely to stick with an option after it helped them avoid losing money (a $0 loss), especially in the stable condition where patterns were easier to detect. This showed that, overall, people could use past outcomes to guide future decisions.

However, a different pattern emerged among individuals with more years of regular substance use. These participants were significantly less likely to repeat a choice that had just helped them avoid a loss. In other words, even when they made a beneficial decision, they often failed to stick with it. Instead of adopting a successful strategy, they tended to switch their choices regardless of whether they had just avoided or incurred a loss, suggesting a broader inconsistency in how they used feedback.

This inconsistency was particularly pronounced in the stable condition, where the best strategy was to recognize and stick with the safer option. Instead, individuals with more extensive substance use histories displayed more erratic behavior, switching choices even when the environment did not warrant it.

To better understand why, researchers used advanced computational models (a Hierarchical Gaussian Filter) to analyze decision-making processes. The analysis revealed that individuals with more years of substance use were less consistent in using “expected value”—the mathematical balance of probabilities and loss magnitudes—to guide their choices. Clinically, this suggests that people with severe substance use histories are not necessarily oblivious to consequences; rather, their brains struggle to consistently apply what they have learned about those consequences to future decisions.

“These results contribute to our conceptualization of substance use severity by suggesting that inconsistency in using cost information, rather than insensitivity to costs, may inform choices to continue using substances despite incurring negative consequences,” Ruiz and colleagues concluded.

The researchers caution that the study has limitations. For instance, the task involved monetary losses rather than real-life substance-related decisions (like drug cues), which may not fully capture the emotional and social pressures people face. Additionally, the study measured cumulative years of use as an indicator of severity, but did not separate current substance users from past users.

The study, “The relationship between regular substance use and cost comparisons in stable and volatile learning contexts,” was authored by Sonia G. Ruiz, Samuel Paskewitz, and Arielle Baskin-Sommers.