Animal Models

Preclinical Models of Addiction: From Self-Administration to Compulsivity

Only about 15–20% of people who use cocaine, heroin, or alcohol will ever develop the compulsive patterns of use that define addiction. The rest use drugs without losing control. This simple epidemiological fact—established in landmark studies by Anthony, Warner, and Kessler (1994)—poses the central question in addiction neuroscience: what biological processes cause some individuals, but not others, to cross the line from recreational use to compulsive drug seeking?

Answering that question in humans is slow, ethically constrained, and confounded by variables that no clinical study can fully control—genetics, environment, polydrug exposure, psychiatric comorbidity. Preclinical animal models exist to solve this problem. By studying drug-taking behavior in genetically diverse laboratory animals under controlled conditions, researchers can isolate the neurobiological mechanisms that drive the transition to addiction and test experimental therapeutics before they ever reach a human subject.

This page is a guide to the major preclinical models used in addiction research today, with a focus on the approaches developed and refined at the Preclinical Addiction Research Consortium (PARC). It covers the foundational paradigms, the recent innovations that bring animal models closer to clinical reality, and the translational pipeline that connects a finding in the laboratory to a potential treatment for patients.

​

The Conceptual Framework: A Three-Stage Addiction Cycle

Modern preclinical models are not designed to recapitulate the entire human experience of addiction. Instead, they target specific behavioral and neurobiological processes within a conceptual framework that has shaped the field for three decades.

The most influential of these frameworks, developed by George F. Koob and Michel Le Moal, describes addiction as a recurring three-stage cycle: binge/intoxication, withdrawal/negative affect, and preoccupation/anticipation (craving). Each stage maps to a distinct domain of brain dysfunction: incentive salience and habit formation in the basal ganglia, negative emotional states in the extended amygdala, and impaired executive function in the prefrontal cortex.

The power of this framework for preclinical research is that each stage can be modeled separately. Self-administration paradigms capture the binge/intoxication stage. Withdrawal and negative affect are studied through measures of anxiety-like behavior, reward thresholds, and stress reactivity. Reinstatement procedures model the preoccupation/anticipation stage by testing what triggers relapse-like behavior after a period of abstinence.

Critically, the framework predicts that addiction involves a shift from positive reinforcement (taking drugs because they feel good) to negative reinforcement (taking drugs to escape feeling bad). This shift from impulsivity to compulsivity is the neurobiological signature of addiction, and modeling it in animals has become the defining challenge of the field.

​

Drug Self-Administration: The Foundation

The drug self-administration paradigm is the cornerstone of preclinical addiction research. In its simplest form, an animal is surgically implanted with an intravenous catheter and placed in an operant chamber where pressing a lever delivers a dose of drug directly into the bloodstream. If the animal presses the lever repeatedly, the drug is considered reinforcing.

This basic setup, refined over decades, can be configured in numerous ways to model different aspects of drug-taking behavior:

Fixed-ratio schedules require a set number of lever presses per drug infusion (e.g., one press = one infusion under an FR1 schedule). These are used primarily during acquisition—they establish that an animal will self-administer a drug and allow researchers to measure baseline intake.

Progressive-ratio schedules increase the number of presses required for each successive infusion (e.g., 1, 2, 4, 6, 9, 12, and so on). The point at which the animal stops pressing—the breakpoint—is a direct measure of how motivated the animal is to obtain the drug. High breakpoints indicate high motivation, a key feature of addiction.

Second-order schedules introduce conditioned stimuli (a light or tone paired with drug delivery) that maintain drug-seeking behavior over extended periods. These schedules model the sustained, cue-driven seeking behavior that characterizes human addiction between drug-taking episodes.

Self-administration has been adapted to virtually every drug of abuse (cocaine, heroin, fentanyl, methamphetamine, nicotine, and alcohol) as well as to different routes of administration, including oral, intra-pulmonary (modeling smoking/vaping), and intragastric delivery. PARC operates one of the largest alcohol and drug self-administration programs in the United States, running multiple drug cohorts simultaneously across standardized protocols.

​

The Escalation Model: From Controlled Use to Compulsive Intake

A pivotal advance came in the late 1990s and early 2000s with the development of the extended-access, or escalation, model. The key insight was simple but transformative: animals given only brief daily access to a drug (typically 1–2 hours, called short access or ShA) maintain stable, controlled intake levels. Animals given extended daily access (6 hours or more, called long access or LgA) gradually escalate their intake over days and weeks, taking progressively more drug per session.

This escalation of intake is not just a matter of taking more drug. Long-access animals also show elevated breakpoints on progressive-ratio schedules, indicating increased motivation. They continue to self-administer despite negative consequences, such as pairing drug infusions with a mild electric footshock. And they exhibit greater seeking behavior after a period of forced abstinence.

In short, extended access produces animals that look, behaviorally, like individuals with a substance use disorder, while short-access animals look like controlled recreational users. The escalation model thus recapitulates the clinical observation that prolonged, heavy exposure increases the risk of transitioning to addiction, and it provides a within-experiment comparison between controlled and compulsive-like phenotypes.

​

The Addiction Index: Quantifying Compulsivity

The escalation model demonstrated that prolonged exposure changes behavior, but it still treated animals as groups. The next critical step was recognizing that, just as in humans, individual animals within the same cohort vary dramatically in their vulnerability to compulsive drug use.

To capture this variation, PARC developed the Addiction Index (based on the work of Deroche-Gamonet et al., 2004), a quantitative framework for classifying individual animals along a compulsivity continuum based on three behavioral criteria that parallel the clinical diagnostic criteria for substance use disorder:

1. Escalation of intake: Does the animal progressively increase drug consumption beyond baseline levels, indicating loss of control over intake?

2. Increased motivation: Does the animal show elevated breakpoints on progressive-ratio schedules, indicating it will work harder to obtain the drug?

3. Continued use despite adverse consequences: Does the animal persist in drug self-administration even when each infusion is paired with a mild punishment (electric footshock)?

Each animal receives a score on each criterion, and the composite Addiction Index places it on a continuous spectrum from resilient (low scores on all three) to highly vulnerable (high scores on all three). In typical cohorts of outbred heterogeneous stock rats, approximately 15–20% of animals meet criteria for high compulsivity—a figure that closely mirrors the epidemiological data on human addiction prevalence.

The Addiction Index is not merely a classification tool. It is the foundation for PARC’s biobank program: every biological sample in the Cocaine Biobank and Oxycodone Biobank is linked to the animal’s Addiction Index score, allowing researchers to compare the neurobiology, genetics, and molecular profiles of vulnerable versus resilient individuals. This approach transforms a tissue bank from a collection of anonymous samples into a precision resource for studying the biological mechanisms of addiction vulnerability.

​

Beyond Self-Administration: Complementary Models

Drug self-administration captures the core of addiction, voluntary and compulsive drug-taking, but several complementary paradigms address dimensions that self-administration alone cannot fully model.

Reinstatement procedures model relapse. After an animal has undergone self-administration and then a period of extinction (during which lever pressing no longer delivers drug), relapse-like behavior is triggered by one of three stimuli: a small priming dose of the drug, exposure to cues previously associated with drug delivery (lights, tones), or acute stress. Each trigger engages different neurocircuitry, and the reinstatement model has been instrumental in identifying pharmacological targets for relapse prevention.

Conditioned place preference (CPP) measures the rewarding properties of a drug by assessing whether an animal spends more time in an environment paired with drug exposure versus a neutral environment. CPP does not require surgical catheter implantation, making it faster and suitable for large-scale screening. It is particularly useful for studying the conditioned associations that drive drug-seeking behavior.

Intracranial self-stimulation (ICSS) measures the brain’s reward threshold. Animals with electrodes in reward circuits will press a lever to deliver a small electrical stimulation. During withdrawal from drugs, the reward threshold rises—meaning the animal needs more stimulation to achieve the same level of reward. This elevation provides a direct measure of the anhedonia and negative affect that characterize the withdrawal/negative affect stage of addiction.

Each of these paradigms contributes a piece of the puzzle. PARC’s approach is to deploy them in combination, building a multi-dimensional behavioral profile for each animal that captures motivation, reward sensitivity, cue reactivity, stress responsiveness, and compulsivity.

​

Individual Differences: The Compulsivity Continuum

Perhaps the most important conceptual shift in preclinical addiction research over the past two decades has been the move from group-level analyses to individual-level phenotyping.

Historically, preclinical studies reported group means: the average intake of a cohort of animals given long access to cocaine, the average reinstatement response after a stress prime. This approach obscured the most clinically relevant observation—that within any cohort, the majority of animals maintain controlled use, while a vulnerable minority develops compulsive-like patterns.

By using outbred heterogeneous stock (HS) rats, which carry broad genetic diversity analogous to a human population, and phenotyping each animal on the Addiction Index, PARC’s work has shown that vulnerability to compulsive drug use is not randomly distributed. It clusters with specific behavioral traits (high novelty-seeking, high impulsivity), specific neural network signatures (altered connectivity in prefrontal-striatal-amygdalar circuits), and specific genetic variants (identified through genome-wide association studies in the same behaviorally characterized animals).

This individual-differences approach is the bridge between preclinical models and precision medicine. If we can identify the biological factors that determine where an individual falls on the compulsivity continuum, we can, in principle, predict vulnerability in humans and tailor interventions accordingly. 

​

The Translational Pipeline: From Model to Medicine

Preclinical models are not an end in themselves. Their value lies in what they enable: the systematic testing of therapeutic hypotheses before they reach clinical trials.

PARC investigators use the models described above to evaluate experimental therapeutics across three tiers:

Drug screening and proof-of-concept studies test whether a candidate compound reduces drug intake, lowers breakpoints, or blocks reinstatement. PARC runs these studies for academic collaborators and pharmaceutical partners, using standardized protocols across multiple drugs of abuse.

Mechanism-based studies combine pharmacology with neuroscience tools (optogenetics, chemogenetics, fiber photometry, and whole-brain Fos imaging) to identify the neural circuits through which a compound exerts its effects. This level of mechanistic resolution is impossible in human studies and is the unique value proposition of preclinical work.

Pre-IND (Investigational New Drug) studies provide the safety, efficacy, and pharmacokinetic data required for regulatory submissions. PARC has contributed preclinical data packages supporting the development of novel therapeutics including deep brain stimulation protocols for opioid use disorder, nicotine-degrading enzyme therapies, and anti-drug vaccines.

PARC’s current therapeutic pipeline includes projects targeting opioid, cocaine, nicotine, and alcohol use disorders, spanning small-molecule pharmacology, biologics, neuromodulation, and device-based interventions.

​

Where the Field Is Heading

The next generation of preclinical addiction models will integrate three converging advances:

Multi-omics and systems biology. Rather than studying one gene or one brain region at a time, researchers are now combining genomics, transcriptomics, proteomics, and metabolomics with behavioral phenotyping to build comprehensive biological profiles of addiction vulnerability. PARC’s biobanks, with their behaviorally characterized tissue samples from hundreds of genetically diverse animals, provide the substrate for this systems-level approach.

Whole-brain network analysis. Techniques like iDISCO+ tissue clearing and light-sheet microscopy allow researchers to map the activity of every neuron in a rodent brain at single-cell resolution. By applying graph-theoretic network analysis to these whole-brain “brainprints,” PARC has begun to identify how brain networks reorganize during the transition from controlled use to addiction and how different pharmacological interventions restructure those networks.

Translational genetics. Genome-wide association studies in large cohorts of outbred rats are identifying gene variants associated with addiction-related behaviors. By cross-referencing these variants with human genetic databases, we are building the foundation for polygenic risk prediction, the ability to estimate an individual’s genetic predisposition to compulsive drug use based on the combined effects of many small-effect variants.

Together, these advances are moving preclinical addiction research from a descriptive science toward a predictive one. The long-term vision is precision medicine for addiction: matching the right intervention to the right patient based on their individual biological profile. The preclinical models described on this page are the tools that make this vision possible.

 

The Preclinical Addiction Research Consortium (PARC) is a multi-investigator research group based at the University of California, San Diego. PARC develops and applies state-of-the-art animal models to discover the neurobiological mechanisms of addiction and accelerate the development of experimental therapeutics. Learn more about our research programs, explore the Addiction Biobank, or contact us about collaborative opportunities.