Why Some People Keep Using Drugs, Even Knowing That It Ruins Their Lives

If you’re like most people, you’ve been impacted in some way by substance use disorders. You may have had family, friends, or other acquaintances who have struggled with drug addiction. As a result, you may have also seen the significant negative outcomes this can cause: the loss of one’s career, health, and social relationships. Due to this devastating impact, researchers have made major efforts to understand what promotes and maintains substance use despite these negative outcomes.

Toward this goal, our new research at the Laureate Institute for Brain Research has found that there may be important differences in how individuals with substance use disorders learn from negative outcomes [1; 2].

When the outcomes of a decision are different from what we expect, this provides an opportunity to learn. Specifically, we have the chance to update our expectations about what would happen if we made that same decision again. For example, try to think of a recent choice you made where the outcome was worse than you expected. Do you think you’ll make that same choice again? Or do you think you’ll decide differently the next time you’re in a similar situation?

This process of changing our behavior when an outcome is better or worse than expected is called reinforcement learning. Intuitively, this leads us to do the same thing more often when outcomes are better than expected, and to do the same thing less often when outcomes are worse than expected.

Importantly, this learning process can be largely automatic; it does not typically require any mental effort. For example, if an action leads to a positive outcome several times, you may simply feel a stronger desire to make that decision again. In fact, it can often take a lot of mental effort to avoid making that same choice. Similarly, with repeated negative outcomes, you may automatically feel stronger and stronger desires to avoid acting in the ways that caused them.

But these processes may act differently in people with substance use disorders, such as addiction to cocaine, methamphetamines, or opioids. Specifically, substance users in our studies showed slower learning from negative outcomes than did healthy individuals (called a slower “learning rate” [1; 2]). In contrast, they showed a faster learning rate than healthy individuals when outcomes were positive.

In scientific studies, such as those performed in our lab, differences in learning rates are often measured using gambling games. For example, imagine that I showed you three slot machines. Then I told you that each time you choose a slot machine, you will either win or lose $5. You can choose between the three slot machines 20 times in a row, but you do not know which one will win the most often. So, you have to learn by trial and error. Now, pretend you played the first slot machine four times in a row and that it led to a win each time, but when you played it the next two times it led to a loss. What do you do?

You have two options. First, you could keep choosing the same machine. After all, the two recent losses might have just been due to chance. The machine seemed good in the beginning and it might still be the best on average. The other option would be to decide that this machine is probably not the best and to switch to a different one instead.

If you decided to switch to a different machine after the two losses, this would suggest you have a fast learning rate from negative outcomes; in other words, you made a quick change in your behavior after only a couple of bad experiences.

On the other hand, some people might continue to choose the same option five or six more times, even if they continued to lose. This would instead suggest a slow learning rate from negative outcomes.

When we invited several individuals with substance use disorders into our lab and had them play a similar game, they tended to choose the same option more times despite repeated losses. In contrast, healthy participants switched machines more frequently after fewer losses. Those with substance use disorders also showed more randomness in their choices than the healthy participants (e.g., switching away from a winning machine for no reason).

These new results might help us better understand why it can be so easy for some people to develop a substance use disorder and why it can be so hard for them to stop using a drug once they’ve started. For example, it can be very rewarding when first using an addictive drug, which can make us want to keep using it. But these drugs eventually lead to many negative life consequences as well, such as losing one’s friends, family, or career.

Empathy Essential Reads

New Research on Raising Empathetic Children

2 Tools to Improve Any Relationship

You might think these negative consequences would lead a person to stop using drugs, but for some, it simply doesn’t. This might seem puzzling, but it makes sense if the learning rate from negative outcomes is slow. And even if a person does see positive outcomes in treatment, greater randomness in choice could explain why they decide to return to drugs anyway.

How might this information be helpful? First, we might try to find ways in treatment to increase a person’s learning rate from negative outcomes. For example, future studies could try to help people recovering from addiction to pay more attention to negative consequences whenever they decide to use. They could also find ways to ensure that substance users understand that these negative outcomes were caused by drug use and that they don’t blame them on something else instead.

Another possibility is to see whether measuring learning rates can help us predict who will have more difficulty in treatment and who might respond better to some treatments than others. In one of our studies [1], improvements after treatment showed some possible connections to differences in learning rates. Those with greater symptoms a year later also continued to show more randomness in choice. We need to do many more studies to confirm these relationships and examine other treatments, but it is promising to think these measures might provide some useful information about how much a treatment will help a specific person.

It's important to be clear that there’s a lot we still don’t know. For example, it’s possible that substance use causes these changes in learning rates. But it’s also possible that people show these differences in learning rates ahead of time, and that this, instead, makes them more vulnerable to addiction when they begin to use.

Further, while our results might suggest it could help to increase learning rates from negative outcomes, this is no guarantee. It’s also possible, for example, that slower learning from negative outcomes could be helpful in treatment­­. This is because the recovery process can be painful, and people with slower learning rates from these unpleasant experiences may have an easier time deciding to stick with it.

It’s also important to stress that our work is a only small part of a much larger set of studies carried out by other researchers, who have uncovered related patterns of unhealthy decision-making in addiction (if you’d like to read a scientific article summarizing these other studies, see [3]).

Despite the unanswered questions, these results offer some additional insights into the processes that may be involved in substance use disorders and why it can be so difficult for some people to overcome them. They also suggest it might be useful to pay attention to these differences in ourselves.

For example, if something has worked well for you for a long time, but then a couple of bad outcomes occur, do you think you might change your mind too quickly? Or do you instead tend to stay in bad situations too long before deciding to try something different?

Paying attention to these patterns can help us better understand ourselves and potentially identify ways in which we could work toward improving them if needed. They might also help us gain more understanding and empathy for those struggling to overcome problems with substance use.

Facebook/LinkedIn image: Dmytro Zinkevych/Shutterstock