Drugs as instruments: A new framework for non-addictive psychoactive drug use

This is a theoretical and integrative review rather than an empirical trial. The authors synthesise evidence from epidemiology, ethology, experimental pharmacology, neuroscience, and behavioural ecology to build a conceptual model. They assemble published findings on prevalence, subjective motives, animal self-administration, neuropharmacology of major drug classes, and learning and memory research to support and constrain their framework. The methodological logic is to (1) define psychoactive drugs and non-addictive use; (2) catalogue common, empirically reported motives or "instrumentalization goals" for drug consumption; (3) map plausible neuropharmacological mechanisms by which major drug classes can change mental states in ways that serve those goals; (4) introduce a memory taxonomy for drug-related information (declarative and nondeclarative subtypes) that underpins learned instrumentalisation; and (5) propose a two-stage, information-processing model of how drug instrumentalization learning and retrieval operate. The authors also place the model in an evolutionary perspective, arguing that learning mechanisms that originally drove flexible food choice and self-medication plausibly supported drug instrumentalization once psychoactive substances became available in concentrated form. Finally, Miller and colleagues outline empirical and translational implications: suggested experimental paradigms (including animal models that allow animals to use drugs instrumentally to enhance non-drug behaviours), survey instruments and structured interviews to identify instrumentalization profiles in humans, and policy-oriented measures tailored to stages of use. No new primary data or quantitative synthesis (meta-analysis) is reported; the paper develops concepts and testable predictions from existing literature.

The principal output is a multi-component conceptual framework. Central to it is the definition of drug instrumentalization as a two-step learned process: (1) seeking and consuming a psychoactive drug to produce a particular, previously learned mental state; and (2) performing non-drug behaviours under that drug-induced state so as to improve performance and goal attainment. Mental states are conceived as relatively prolonged modes of neural function determined by modulatory transmitter systems (dopamine, serotonin, acetylcholine, noradrenaline, neuropeptides) and their tonic/phasic dynamics. Miller and colleagues enumerate a set of instrumentalization goals, each linked to plausible neuropharmacological mechanisms and drug classes: improved social interaction (alcohol via enhanced GABAergic inhibition and indirect DA increases; low-to-medium-dose psychostimulants via DA/5-HT/NA; NMDA antagonists and GHB at low doses), facilitated sexual behaviour (alcohol, cannabis, amphetamines, cocaine; DA-mediated cue responsivity and conditioned expectancies), improved cognitive performance and counteracting fatigue (caffeine via adenosine receptor antagonism; nicotine via nicotinic ACh receptors; prescription and illicit stimulants via NA and DA systems), recovery from and coping with stress (alcohol, benzodiazepines, barbiturates via GABA A modulation; cannabis via CB1 effects on extinction of aversive memories), self-medication for psychiatric symptoms (diverse classes depending on syndrome), sensory curiosity/expanded perception (classical hallucinogens as 5-HT2A agonists; NMDA antagonists), euphoria/hedonia (opioids and several stimulant drugs with strong mesolimbic DA activation), and improved physical appearance/attractiveness (nicotine and sympathomimetic anorectic effects; anabolic steroids for muscle growth). A key formal advance is a proposed drug-memory taxonomy required for instrumentalization. The declarative drug memory has semantic (facts, expectations) and episodic (autobiographical episodes, context and effects) components. The nondeclarative domain includes classical conditioning, instrumental conditioning, habit memory (stimulus–response control), procedural skills for drug preparation/administration, and drug-priming engrams. Building on these, the authors present a two-stage learning model: Stage I associates contextual stimulus S_y with a drug-taking response R_y that produces a drug-induced mental state DM_s; Stage II shows how a previously learned instrumental action R_x (triggered by stimulus S_x) is executed under DM_s and yields an enhanced outcome. A comparator computes the difference between expected and actual outcomes, providing reinforcement that consolidates both the S_x–R_x and S_y–R_y–DM_s associations. Retrieval requires cues for both the drug-taking context and the target behaviour. The paper discusses life-course establishment: experimentation in adolescence builds semantic and episodic drug memories, and successful self-titration and contextual learning can stabilize controlled instrumentalization. The authors emphasise risks: narrow dose windows for optimal instrumental effects, mis-instrumentalization (taking doses that impair, not help), over-instrumentalization and escalation of intake, and transition to addiction. They review supporting animal findings suggesting animals can regulate drug intake relative to environmental demands, and articulate testable predictions and measurement approaches (structured interviews, questionnaires, and adapted animal paradigms). Finally, the authors set out policy-relevant recommendations: stage-specific education about instrumentalization, screening for instrumentalization patterns in clinical practice, delaying initiation in adolescence, and tailoring prevention to transitional life periods.

Miller and colleagues interpret their synthesis as a reframing: many instances of non-addictive psychoactive use are best understood as learned instrumental behaviours that alter mental states to facilitate goal-directed actions in complex, modern microenvironments. They position this account relative to prior literature by acknowledging established reinforcement and addiction theories while arguing that those accounts pay insufficient attention to stable, functional patterns of non-addictive use. The drug-instrumentalization model is offered as complementary to neuroscience of reward and learning; it emphasises the interplay of multiple transmitter systems, declarative and nondeclarative memories, and context-dependent valuation of mental states. They place the framework in an evolutionary perspective, proposing that species-general learning mechanisms that evolved to guide flexible food choice and self-medication could be co-opted for selecting plants and compounds that change mental states. Modern changes — notably concentrated, purified psychoactives and rapidly shifting microenvironments with diverse behavioural demands — may have increased opportunities for instrumentalization and, for a minority, contributed to maladaptive escalation and addiction. The authors stress, however, that the evolutionary claim is provisional and requires empirical testing. Key limitations and uncertainties are acknowledged. The framework is largely conceptual and based on disparate empirical literatures; direct evidence that instrumentalization increases reproductive fitness is not provided. The authors note interindividual variability (pharmacokinetics, personality, developmental stage) and sociocultural moderators that shape which drugs are available and how they are used. They also highlight adolescence as a high-risk period for establishing instrumentalization and for mis-instrumentalization that may lead to addiction. Methodological caveats include reliance on self-report for many motive data and the need for complementary animal models that allow instrumental use rather than isolated self-administration paradigms. As implications, Miller and colleagues suggest several practical avenues. For prevention and policy they advocate stage-specific education that includes teaching about instrumentalization (not only harms), early screening for instrumentalization patterns in medical settings, delaying initiation in adolescence, and training alternative, non-pharmacological means to alter mental states. For research they recommend developing structured instruments and interviews to characterise instrumentalization profiles, adapting animal paradigms to study instrumental use (where animals may use drugs to enhance non-drug behaviours), and empirically testing the proposed memory taxonomy and two-stage learning model. They call for cautious debate before translating the theory into policy given potential risks and sociocultural complexity.

In closing, Miller and colleagues conclude that non-addictive psychoactive drug use is widespread and often purposive: people learn to use drugs as instruments to alter mental states in ways that can improve performance on survival- and reproduction-relevant goals in modern environments. They reiterate the core contribution — the drug instrumentalization concept, the drug-memory taxonomy, and the two-stage learning model — and argue that a better empirical understanding of instrumentalization could inform prevention and reduce transitions to addiction. The authors emphasise that while instrumentalization may have adaptive utility in many cases, it carries real costs and risks, and careful empirical work is needed to test the framework and to guide any policy responses.