Understanding Addiction: A General Liability or Unique Disorders

Is there a single explanation that accounts for all addictive behaviors, or is the reality more complex? How can quantitative classification methods help uncover the nuances of substance dependence? 

In this episode, APS’s Özge Gürcanlı Fischer Baum speaks with APS Spence Award recipient Ashley Watts from Vanderbilt University about her recent research article in Clinical Psychological Science. They discuss how simple explanations to addictive behavior – like the General Addiction Liability theory – may be appealing for a unified treatment approach, but a deeper, more quantitively rich dive into the data draws a more heterogenous picture. The paper argues that addiction follows a more diverse pattern, and the conversation highlights what this approach to addiction means for future research and for treatment strategies. 

Learn more about Watts’ research and the APS Janet Taylor Spence Award.

Send us your thoughts and questions at underthecortex@psychologicalscience.org.

Unedited Transcript

[00:00:07.640] – APS’s Özge Gürcanlı Fischer Baum

Addictive behaviors often show patterns that connect across different substances. Is there a unitary dimension that links various substance use disorders? Alternatively, do addictive behaviors display a more heterogeneous pattern? Can this novel approach help with understanding certain consequences like social harm and risky behavior? I am Özge Gürcanlı Fischer Baum with the Association for Psychological Science. Today, I have the pleasure of talking to Ashley Watts from Vanderbilt University. Ashley and her colleagues recently published an article on substance use disorders and the general addiction liability theory in APIS’s journal, Clinical Psychological Science. Join us as we discuss the importance of a quantitative approach to addiction across substances like alcohol, cannabis, with tobacco and opioids. Ashley, welcome to Under the Cortex.

[00:01:05.810] – Ashley Watts

Hi, thanks so much for having me.

[00:01:07.940] – APS’s Özge Gürcanlı Fischer Baum

Let me start with our first question. What type of psychologist are you?

[00:01:12.400] – Ashley Watts

To me, a bit of a complicated question. My training is in clinical psychology. In graduate school, I focused largely on the classification of personality disorders. And toward the end of graduate school, I became increasingly interested in quantitative methods and the ways that quantitative methods can be used to improve psychiatric classification. I shifted my focus into the space of addiction on my post-doc. I spent five years at University of Missouri as a postdoc and research professor, where I moved away from studying classification of personality disorders and toward psychopathology more generally. I think of myself as a quantitative psychopathologist, though my personality training creeps in from from time to time, too.

[00:02:01.230] – APS’s Özge Gürcanlı Fischer Baum

It is a unique approach, definitely. That’s why this article is so great, and we are so excited to have you here. What got you interested in studying addiction? How did you shift from studying personality disorders to addiction?

[00:02:15.690] – Ashley Watts

Gosh. My graduate research made the point to the extent possible that we really can’t think of any given personality disorder as one thing. They’re remarkably heterogeneous. They contain many different dimensions that in concert produce what looks like a psychopathic personality, for instance. I knew I wanted to shift away from studying that, and so I took on a post-doc that would challenge me to learn something new. Immediately, I was struck by the heterogeneous nature of substance use disorders. I mean, even just week one of my post-doc, I was like, Whoa, how can we take this collection of symptoms and call this one thing? There’s no way that could be true. I think I projected a little bit of my interest in classification in heterogeneity onto addiction and hit the ground running by focusing on sources of heterogeneity in addiction.

[00:03:09.290] – APS’s Özge Gürcanlı Fischer Baum

Yeah. Every case is unique. That is the perspective I think you are you from and you talk about the general addiction liability in your study, we will come to that. But let’s talk about the landscape of addiction in general. What is the landscape of addiction and substance use disorders like in the United States?

[00:03:33.260] – Ashley Watts

I think our current landscape is both exciting and scary at the same time. We’re in the throes of the fourth wave, as people are calling it, of the opioid epidemic. We’ve witnessed a tragic loss from the use of opioids, then fentanyl, heroine, and now it’s evolving also into stimulant use. This is a really scary time to be a substance use researcher, and yet also I see a lot of potential for improving the way that we classify addiction and improving our ability to get people access to the treatments that we know can help them. When I say exciting, too, in addition to having an optimistic perspective that improving classification can indeed help people, we’re also in this period of cultural, sociopolitical change in terms of the way that we view drug use and the people who use drugs. With cannabis being a great example, we have this experiment going on in our country where we’re decriminalizing and legalizing a schedule one substance, the most criminalized substance category in the country. As our attitudes at a broad, if not federal level, at least at general state levels, are changing, it’s also shifting the way that we conceptualize a substance use disorder.

[00:04:53.840] – Ashley Watts

I think these are interesting times because as we are adopting more flexible views use, politically speaking, toward drugs, it’s also shifting the way that we view people who use drugs and in turn classify drug use.

[00:05:09.110] – APS’s Özge Gürcanlı Fischer Baum

Yeah. Definitely, the way we classify drug use has clear implications for policy and interactions and other further steps. Yeah, thank you for that. Let’s talk about the general addiction liability theory. What is it and why is this an idea that people are excited about?

[00:05:29.110] – Ashley Watts

General addiction liability, the general concept has been around for, I’d say, 15 or so years, but the field of clinical science, psychiatry included, has become a little enamored with general factors in general. Really, everyone is trying to describe really remarkably varied or heterogeneous forms of psychopathology in terms of unitary dimensions. General addiction liability, the idea behind it is that it reflects or captures a person’s liability to experience experience any form of addiction. For the most part, people have focused on drug addiction, though some people view addiction as potentially being expanded into the behavioral realm to include things like gambling, shopping, sex, and so on. But Most of the literature is focused on this idea that if you use one drug, it’s because you have a propensity to experience addiction in general, and that will increase the likelihood that you will become addicted to two, three, four, and so on drugs. That’s the general idea behind it, but it’s very much influenced by quantitative methods and the fact that we use certain kinds of quantitative methods that are, I would argue, slightly biased in terms of producing statistical evidence for these constructs. In my view, quantitative methods have encouraged the study of these general factors more than strong theory.

[00:06:56.100] – APS’s Özge Gürcanlı Fischer Baum

Basically, the idea is, can all these I guess different experiences boil down to one general way of being, which is in public, we talk about it as addictive personality. Does this person have an addictive personality? Totally. I see it a lot on social media recently, too. But we need to use the right data, the right way to classify these addiction behaviors. I will come to that. But I am also curious about why people are excited about this type of concept. If there were to be strong support for the general addiction liability theory, what would that practically mean for diagnosis and how practitioners think about addiction?

[00:07:48.740] – Ashley Watts

If general addiction liability were a real thing, I can understand why it appeals to scientists and potentially practitioners. Because what it means is that you can quantify a person’s liability toward addiction based on one score. You can have one piece of numerical data for someone, and it tells you their risk of developing addiction for any drug. That’s appealing because it’s a simple explanation. But unfortunately, things are not that simple. For practitioners, what it would mean then is that there’s just rampant comorbidity among substance use disorders. If you have someone who presents with an alcohol use disorder, it means they probably also have a cannabis use disorder, maybe also have a tobacco use disorder, and so on. The data are just not on that side. That’s a good thing. It’s good that people don’t present with 7, 8, 9 drug use disorder diagnosis. That’s extremely exceedingly rare. But that’s what that would mean if that were the case. In fact, we see that that’s not true. Most people do not diagnose with a substance use disorder at all. If they do, they might have one or two. Beyond two, That’s just statistically what one of my old advisors would call decimal dust, meaning it’s just not represented in the population. We’re talking about less than 1% of people.

[00:09:10.900] – APS’s Özge Gürcanlı Fischer Baum

Yeah. Thank you for that. I think you, in a way, answered my next question, but I will ask it anyway, if you don’t mind, so you can elaborate. Why did you think going into this study that there might not be widespread support for general addiction liability?

[00:09:28.520] – Ashley Watts

That’s a great question. I work with a lot of large scale population level and epidemiologic data. This data are fun to work with because the samples are well-powered to detect interesting population-wide effects, and at the same time, they’re pretty tricky to work with. In the case of substance use disorder, again, most people do not diagnose with a substance use disorder. In epidemiologic data, we’re talking about, say, 80% to 90% of people who do not achieve a single substance use disorder diagnosis. Then when you’re looking at what glues many substance use disorders together, if you are also including folks who don’t use drugs at all, it becomes complicated because what you’re talking about is really a lack of substance use at the population level rather than the presence of numerous substance use disorder diagnosis. My thinking is, okay, well, what we’re really talking about and what we’re quantifying at the population level is a lack of substance use rather than addiction liability. That seems potentially counterintuitive to some people, but if we’re talking about a group of people who maybe use one drug, maybe two, there’s no way we can really quantify shared liability toward addiction because people aren’t necessarily exposed to the myriad forms of drugs that would be necessary to quantify that liability.

[00:10:57.240] – Ashley Watts

I became interested in it only because my familiarity with these sorts of data and what they can actually tell us.

[00:11:06.240] – APS’s Özge Gürcanlı Fischer Baum

Yeah. The results are interesting and important for our field. What did you end up finding?

[00:11:15.960] – Ashley Watts

I would argue with some very low support for general addiction liability. It’s the case that if you take four substance use disorder diagnosis and just model them as binary, yes, no, does this person have a diagnosis? If you extract the common or shared variants among those, you can find some factor, some latent variable. But once you start subjecting that research question to riskier tests or more scrutinous tests, the data falls apart pretty quickly. If you model, for instance, symptoms as opposed to binary diagnosis, we see that the symptoms just don’t hang together all that well. If you focus on within-person analysis, which we did, too, you don’t see evidence that a large majority of people in the population have high levels of more than one substance use disorder. We conducted a variety of different kinds of tests with a variety of quantitative methods to really dig into the data to see where the story might be. We came up short in terms of supporting general addiction liability.

[00:12:21.260] – APS’s Özge Gürcanlı Fischer Baum

Yeah. Let’s go into the details of that a little bit because methodologically, I really like this article because it asked a simple question, do we find support for this theory? Like you just said, you use various methods. The main study you cite for general addiction liability, you analysis one model and overall low specificity items for the factor analysis. Do you think a factor analysis of this nature could be asking a different question than models that look at highly specific items and the relations between them?

[00:13:02.830] – Ashley Watts

I think so, especially as I said a minute ago, if you just take four diagnosis and you extract a factor that reflects them, when most people do to have comorbid SUD. We’re really talking about people who have all zeros, which is a good thing. It means they don’t have any diagnosis of substance use disorders. But what glues together a lack of substance use disorder is not necessarily what glues together presence of substance use disorder. Because in the case of a drug use disorder, you have to use the drug to even become addicted to it. You’ve not been exposed to what is necessary to achieve a diagnosis of substance use disorder. We’re probably really talking about mechanisms that are involved in a lack of drug consumption in general, or a lack of comorbidity rather than the presence of comorbidity. That’s, I think, a little challenging for some people, including myself for years, to wrap your head around. But the The sense of comorbidity is not equivalent to the lack of comorbidity of drugs or any psychopathology, for that matter. They could involve completely distinct sets of processes, with drug consumption being one that clearly differentiates those two groups.

[00:14:15.320] – Ashley Watts

I do think that we’re getting not only statistically, but conceptually at something very different when we just toss four diagnosis into a model. It’s also worth mentioning that those binary diagnosis tell us very little because of how heterogeneous a given substance use disorder is. I mean, they’re remarkably varied in terms of the profiles of symptoms that people endorse. We gloss over just massive amounts of information when we give people a zero or a one on a substance use disorder. There’s so much more under the hood there in terms of the symptoms that they’re endorsing, what configurations of symptoms people have. It’s just a remarkably simplistic approach to an extremely complicated set of phenomena.

[00:15:02.170] – APS’s Özge Gürcanlı Fischer Baum

Yeah. I mean, this is not black and right. It is not that simple, and you make it very clear with your article. I would like to talk a little bit about consequences of substance use. We have been talking, it is interesting that despite having relatively little overlap in symptomology relationships across substance use disorders, we have been talking about this very heterogeneous nature of things. Although there were strong relationships across substance use disorders on the consequences of substance use. How do you think that might affect how practitioners might consider treating?

[00:15:44.230] – Ashley Watts

I love this question. I love consequences. I have evolved in terms of my thinking on consequences. To give readers, or listeners rather, some background, if they’re not overly familiar with the In the context of substance use disorder diagnosis, there are four or five diagnostic criteria in substance use disorders that capture things that have gone wrong in your life that are potentially associated with your use of a drug. For instance, there’s a criterion called hazardous use. In alcohol, for instance, that generally indicates that you are engaging in drinking and driving. The idea is you are engaging in recurrent patterns of risky behavior that are downstream or caused by the drug. There are several examples of criteria like this in substance use disorder. The idea is that if you are using a drug and you’re getting into trouble, it must be that you’re engaging in a compulsive pattern of use. Because why would you continue using a drug if you keep getting in trouble? The idea is if you’re using in the face of punishment, you’re inherently a compulsive drug user. That’s the theoretical model or framework that’s guided diagnosis. The issue is that there are certain types of people that are prone to consequences, regardless of whether or not they are drinking or using drugs.

[00:17:11.720] – Ashley Watts

There are externalizing prone people, for instance, people who are inherently risk-taking or impulsive, those folks could engage in risky behavior when they’re drinking and when they’re not drinking. From a diagnosis perspective, we are baking in variants related to constructs that are not related to addiction. What I mean is we are assuming that those criteria are endorsed by people who engage in compulsive drug use, but also those same criteria are endorsed by externalizing drug people who get themselves into trouble regardless of their drug use data. From a diagnosis perspective, we are probably overdiagnosing impulsive people who take a lot of risk, simply because we baked in that exact configuration of psychopathology and personality traits.

[00:18:09.980] – APS’s Özge Gürcanlı Fischer Baum

Yeah, that is really interesting. Do you have anything else to add to your explanation?

[00:18:17.150] – Ashley Watts

Sure. We think of these criteria as capture and compulsion. They probably most of the time reflect impulsivity. What that means is that we’re probably overdiagnosing, externalizing prone people with substance use disorders, and we’re not capturing the folks that we intend to capture, the people who are truly so-called addicted to a given drug. As an example of my argument, there’s an old psychopathy theorist and clinician Hervie Cleckley from the 1940s, he wrote this book called The Mask of Sanity. He talked about the psychopath in a way that exemplifies this exact point. He said the psychopath is prone to fantastic and uninviting behavior with drink and sometimes without. The point there is when psychopaths are drunk, they’re reckless. They engage in lots of antisocial and aggressive behavior. They antagonize other people. But guess what? They also do that when they’re sober. In the context of substance use disorder, what we see is that the people that are getting kicked out of bars, maybe they’re antagonizing people, getting into fights, coming home to their spouse and causing an argument. That It might not be addiction, but it might reflect the specific externalizing relevant manifestation of drug use. My work is trying to tease apart those folks from the people that we flag as, We think you’re truly, truly addicted to a drug.

[00:19:46.500] – Ashley Watts

That phenotype, that externalizing, rowdy, heatless, drinking, drug use thing is interesting clinically, but it doesn’t mean that that person experiences addiction. I’m trying to tease apart those two phenotypes, which I think are dramatically different in terms of the implications for classification and treatment.

[00:20:08.100] – APS’s Özge Gürcanlı Fischer Baum

Yeah. Just like in a diagnosis, comorbidity is always an issue, and you described it really Well, thank you very much for that. Let’s go back to the general addiction liability theory. If evidence continues to support for this theory, what would the implications What does it mean for how we think about substance use disorders and how we think about treatment? Where do we go from here?

[00:20:39.310] – Ashley Watts

If general addiction liability turns out to be a robust, well-supported phenomenon I think it will cause us to shift the way that we treat folks. Right now, for the most part, we put people in drug-specific treatment programs. Of course, there are some transdiagnostic elements sense of treating addiction. But to some extent, because of the unique pharmacologic and physiologic properties of drugs, we tend to focus on the unique characteristics of a drug in treatment. With methadone, for instance, being a way to taper people off of prescription opioids, as just one example. That’s not something you would necessarily administer if someone is experiencing withdrawal from alcohol. We do have drug-specific treatments, especially when we’re thinking of medicine rather than psychological treatment. But if a general addiction liability is a robust phenomenon, in addition to the unique pharmacologic targets for a given drug use disorder, it probably means that we have to shift our treatment toward transdiagnostic approaches. In other realms of clinical psychology, we’ve moved toward targeting broad elements like a general tendency to experience negative affect. Now there are transdiagnostic treatments that target specifically this general tendency to have negative affect. In the same way, we could develop treatments that target whatever is captured or quantified in this general addiction liability dimension or more, and we could target that.

[00:22:18.080] – Ashley Watts

The problem is, I don’t know what that is at this time. Is it impulsivity? If so, we have some treatments that can work for impulsivity. But again, I think that’s slightly separate from addiction per se.

[00:22:30.550] – APS’s Özge Gürcanlı Fischer Baum

Yeah, that’s a great question. I think there is so much to be done in the field, and studies like yours can contribute with data support and a quantitative perspective. Ashley, thank you so much. This was a great conversation, and I personally learned a lot. Thanks again for joining us Under the Cortex.

[00:22:53.960] – Ashley Watts

Thank you so much.

[00:22:55.920] – APS’s Özge Gürcanlı Fischer Baum

This is Özge Gürcanlı Fischer Baum with APS, and I have been speaking to Ashley Watts from Vanderbilt University. If you want to know more about this research, visit psychologicalscience.org. Do you have questions or suggestions for us? Please contact us at underthecortex@psychologicalscience.org.