Psychology by BogdanovaThe Encyclopedia of Psychology for the Artificial Era

RESEARCH DOSSIER

Addiction-Related Attention, Executive Control, and Decision Processes

Abstract. Addiction-related cognition arises from interactions among conditioned attentional priorities, executive regulation, learned action tendencies, and changing valuation of immediate and future outcomes. This fundamental research monograph critically synthesizes historical experimental paradigms, incentive-sensitization, habit and impaired response inhibition and salience attribution models, human neuropsychological studies, behavioral meta-analyses, decision paradigms, and clinical training trials through 2026. It separates go/no-go withholding from stop-signal cancellation, attention-bias effects from individual prediction, and experimental change from clinically meaningful treatment outcomes. Evidence indicates substance- and context-dependent average differences, modest association between attentional bias and craving, variable prospective prediction, psychometric limitations of several task contrasts, and conditional or inconsistent clinical transfer of cognitive retraining. The research emphasizes confounding, measurement reliability, state changes, heterogeneity, ecological validation and qualified scientific conclusions. It presents an original critical narrative synthesis of published research without original participant data, a preregistered systematic review, or an independent meta-analysis.

Keywords: addiction psychology; attentional bias; addiction Stroop; visual-probe task; eye tracking; approach bias; cue reactivity; craving; response inhibition; go/no-go task; stop-signal task; executive function; working memory; cognitive control; decision-making; Iowa Gambling Task; delay discounting; habit formation; cognitive bias modification; inhibitory control training; relapse prediction; ecological validity; Artificial Era

Introduction: Addiction as a Problem of Contextual Cognitive Regulation

Addiction-related attention, executive control, and decision processes concern the ways substance-associated information competes for perceptual priority, modifies action readiness, and changes how short-term outcomes are evaluated against alternative goals. The research object is a family of interacting processes rather than a single deficiency. Attention can be drawn toward a conditioned cue while a person also retains long-term intentions to avoid use; conflict arises when these tendencies must be regulated under time pressure, stress, intoxication, deprivation, or changing environmental opportunity. This monograph examines how those processes are experimentally measured, what can reasonably be inferred about their causal contribution, and which claims have survived clinical and methodological scrutiny Field & Cox (2008) Goldstein & Volkow (2011) Ceceli et al. (2026).

The guiding questions concern precedence, mechanism, specificity, and prediction. Do attentional and inhibitory differences precede harmful exposure or partly follow it? Does an addiction-related cue change cognition above and beyond generic attention or response speed? Can a model explain how executive functions operate in conjunction with valuation, craving, habit, and alternative reward? Which measurements predict behavior outside the laboratory, and when do laboratory changes translate into durable clinical benefits? The evidence base comprises original cognitive and neuropsychological experiments, patient-control comparisons, longitudinal observations, trials of cognitive-bias interventions, systematic reviews, and measurement-method studies. The methodological approach is a critical narrative synthesis rather than an original experiment, preregistered systematic review, or meta-analysis Smith et al. (2014) Hedge et al. (2018) Pan et al. (2026).

An adequate explanation must allow individual change. Both intense drug-cue reactivity and good inhibitory performance can occur in the same person at different times, while favorable control-task scores may coexist with considerable clinical risk when intoxication, withdrawal, social contingencies, or immediate access change. Group differences provide a population-level signal; diagnosis and individualized prediction require independent validity. Accordingly, this study follows the entire inferential pathway from experimentally defined construct through measurement and model to meaningful action and recovery Liu et al. (2019) Schulte et al. (2014) Vafaie & Kober (2022).

Scientific Scope, Definitions, and Diagnostic Boundaries

Attentional bias refers to a systematic difference in the allocation, capture, maintenance, or disengagement of attention as a function of stimulus relevance. In addiction research the relevant contrast usually concerns alcohol-, tobacco-, drug-, or gambling-associated material against comparison material selected to control perceptual and affective characteristics. Inhibitory control denotes the capacity, under specified task conditions, to withhold or interrupt a response; executive control more broadly includes maintaining goals, shifting strategies, updating working memory, monitoring conflict, and organizing action in the service of task requirements. Decision processes describe how options are represented, valued, selected, and revised after outcomes Field & Cox (2008) Smith et al. (2014) Verbruggen et al. (2019).

These definitions concern operational constructs, not diagnostic entities. Substance use disorder is established by a recognized clinical pattern of difficulties and impairment in context; no dot-probe score, stop-signal reaction time, brain scan, or Iowa Gambling Task result can establish it independently. Tolerance and withdrawal are physiological and clinical phenomena whose interpretation depends on substance and history. Craving is a motivational state or report, and impulsivity comprises multiple dispositions and behaviors. The research therefore distinguishes established disorders and relevant behavioral addictions from task-defined phenomena that may or may not be clinically informative Goldstein & Volkow (2011) Ceceli et al. (2026).

The dossier investigates human substance-related research primarily and draws on gambling-disorder findings where the experimental question genuinely transfers. It examines exposure, acute intoxication, withdrawal, prolonged abstinence, and recovery as distinct contexts. General theories of attention and inhibition receive only the development necessary to evaluate addiction-specific findings; their wider scientific treatment belongs to the existing Attention and Action Selection and Inhibition dossiers. The analysis of intertemporal choice remains focused on its relationship to attention and control rather than duplicating the specialized Delay Discounting dossier MacKillop et al. (2011) Peixoto et al. (2025).

Historical Formation of Cognitive Addiction Research

Experimental psychology developed selective-attention paradigms, response-competition procedures, and decision tasks before these tools entered addiction science. Classical interference research showed that task-irrelevant material could delay an instructed response, establishing a methodological basis for investigating the processing priority of personally meaningful cues. The subsequent adaptation of Stroop-like procedures to alcohol and drug words offered a striking and replicable research question: might substance-associated information capture processing resources despite an unrelated task? Field and Cox's historical and theoretical review places these adaptations alongside visual-probe, eye-movement, and conditioning approaches Field & Cox (2008).

A second tradition developed from the measurement of action selection and stopping. Go/no-go and stop-signal tasks sought to distinguish omission of a prepotent response from the cancellation of a response already under preparation. In parallel, neuropsychological studies investigated altered choice on decision tasks following prefrontal lesions. Bechara and colleagues' 1994 lesion work and their 2001 comparison of individuals with alcohol or stimulant histories to controls helped establish questions about future consequences and decision performance, though the designs do not imply that addiction reproduces the causal effects of focal lesions Bechara et al. (1994) Bechara et al. (2001) Verbruggen et al. (2019).

These traditions were joined by influential addiction models. Incentive sensitization predicted motivational amplification of drug cues; habit-centered accounts examined the shift from goal-directed action toward stimulus-driven responding; the impaired response inhibition and salience attribution model emphasized interactions between prefrontal control and drug-cue significance. The literature's history consequently reflects several partially compatible levels of explanation, from conditioned salience to action learning to executive resource allocation. Their integration must be tested with tasks sensitive to the interactions they describe Robinson & Berridge (1993) Everitt & Robbins (2016) Goldstein & Volkow (2011) Ceceli et al. (2026).

Conditioned Cues, Incentive Salience, and Attentional Priority

Repeated associations between an action, a substance, and its predictors can give previously ordinary stimuli the capacity to elicit anticipation and approach. Incentive-sensitization theory specifies a process by which cues acquire motivational pull that can diverge from the hedonic pleasure obtained through consumption. Thus a familiar place, object, smell, or social context may alter where attention is directed and which action becomes immediately compelling. The distinction between motivational wanting and hedonic liking is essential: an attentional preference is not simply evidence of pleasure, and subjective craving is an imperfect indicator of the underlying motivational processes Robinson & Berridge (1993) Robinson & Berridge (2025) Berridge & Robinson (2016).

Stimulus salience is conditional. Drug-related material can be more or less attractive depending on abstinence, expectations, availability, current goals, competing incentives, and the likelihood that an action will be rewarded. Evidence from visual attention tasks should therefore identify the stage at which a cue matters: initial orienting, sustained gaze, difficulty disengaging, or later strategic attention. These mechanisms can yield superficially similar group averages while making different predictions about timing and intervention Field & Cox (2008) Field et al. (2009).

A critical research priority is separating associative learning from motivational causation. A photograph can capture attention because it is visually distinctive, familiar, emotionally charged, or explicitly relevant to the task. Designing matched stimulus sets and manipulating learned contingencies can limit these alternatives, but a significant drug-versus-neutral contrast alone does not prove sensitized neural incentive systems. The strongest inferential strategy combines learning histories, temporal measurements, experimental manipulation, and out-of-sample prediction Robinson & Berridge (2025) Vafaie & Kober (2022).

The Addiction Stroop: Interference and Interpretive Ambiguity

An addiction Stroop task typically requires participants to identify the ink color of substance-related and comparison words while ignoring word meaning. Longer response times or higher error rates on substance words are interpreted as interference from processing meaning that competes with the instructed task. However, the observed difference is a behavioral composite. Word frequency, emotional valence, semantic associations, response strategies, language proficiency, fatigue, and practice effects can affect it. The task estimates interference under a chosen design; it cannot directly identify a unique attentional subsystem or establish compulsive behavior Field & Cox (2008).

Trial-level processes introduce further complexity. A substance word might draw initial processing resources, encourage extended reflection after color recognition, or influence subsequent trials through carryover. Randomized and blocked task arrangements therefore need distinct interpretations. A block effect can reflect the lingering mood or expectation induced by several related items, while a single trial can be influenced by the immediately preceding stimulus. Reports should specify item matching, trial order, response-time cleaning, error handling, and reliability rather than present a single interference score as a fixed trait Field & Cox (2008) Hedge et al. (2018).

The historical influence of the addiction Stroop exceeds its individual diagnostic value. It revealed experimentally tractable interference associated with substance relevance and stimulated research on cue salience and craving. Yet a cognitive marker useful for studying group tendencies is not necessarily reliable for distinguishing individual patients or predicting treatment response. The reliability paradox explains why tasks engineered to yield robust experimental contrasts may produce weak stable between-person measures Hedge et al. (2018) Enkavi et al. (2019).

Visual-Probe, Spatial-Cueing, and Eye-Movement Paradigms

In the visual-probe task, pairs of addiction-related and comparison stimuli appear briefly before a target replaces one stimulus location. Faster responding when the target appears in the addiction-cue location is commonly treated as evidence of attention bias. Spatial-cueing variants use valid and invalid targets to investigate orienting and disengagement. These paradigms allow systematic manipulation of exposure duration, but reaction times are affected by motor readiness, eye position, anticipatory strategies, and differences in processing speed. A reaction-time contrast must be interpreted relative to the task's assumptions and the reliability of the resulting score Field & Cox (2008) Hedge et al. (2018).

Eye tracking provides a complementary observation by measuring gaze direction, fixation duration, first fixation, and dwell time. It can separate some early and sustained viewing patterns more directly than reaction time alone. Nevertheless, fixation is not identical to covert attention, and eye movements can be influenced by instruction, salience, and laboratory novelty. Stimulus sets need to resemble relevant real-world cues while maintaining comparability, a tension that prevents a universal standardized task from capturing every substance and cultural context Field & Cox (2008).

The association between attentional bias and craving was quantified in Field, Munafò, and Franken's meta-analysis across 68 independent data sets. The reported overall correlation was modest, approximately r = .19. This finding supports some systematic covariance while leaving most variance unexplained. A small correlation is compatible with diverse causal directions: craving may increase cue processing, cue processing may change craving, or both may reflect an underlying state such as expected access. Longitudinal designs and selective experimental manipulations are needed to adjudicate among these accounts Field et al. (2009).

Approach Bias and Automatic Action Tendencies

Approach bias is an action tendency favoring movement or response mappings associated with acquiring a substance-related stimulus. In approach–avoidance tasks, participants may pull or push a joystick in response to images while ignoring their semantic content. A faster pull response to alcohol pictures, relative to avoidance responses, is often interpreted as an automatic approach tendency. But task instructions, response geometry, stimulus compatibility, and general motor asymmetries can create a similar difference. The inference concerns relative action readiness under a specified task rather than an uncontrollable disposition Wiers et al. (2007) Wiers et al. (2011).

Learning histories can render substance cues both attention grabbing and action facilitating. Such coupling might arise because the same cue represents an expected reward, because selecting it has been practiced, or because attention and action share a current goal. Attention and approach bias should therefore be measured separately where possible and modeled as potentially interacting processes. One cannot conclude that a gaze difference entails an approach movement or that approach bias necessarily follows from the amount consumed Field & Cox (2008) Everitt & Robbins (2016).

The practical importance of the construct comes from the possibility of modifying stimulus–action relationships through repeated training. The 2011 randomized work by Wiers and colleagues in alcohol-dependent patients investigated whether altered response contingencies could change approach bias and subsequent outcomes. Its findings motivate intervention research, but any claim of clinical benefit must rest on replicated patient outcomes and well-controlled comparisons rather than successful training on the same task alone Wiers et al. (2011) Pan et al. (2026).

Craving, Cue Reactivity, and Attention–Motivation Coupling

Craving is typically measured through self-report ratings, while cue reactivity may include subjective, physiological, behavioral, or neural responses to addiction-related stimuli. These measurements answer related but distinct questions. A person can report an urge without looking longer at a cue, and an eye-tracked preference can be present in the absence of intense reported desire. Differences in timing, access expectations, abstinence, and the content of a cue can change the relationship among them. Theoretical integration therefore requires explicit measurement of both response and context Field et al. (2009) Robinson & Berridge (2025).

Vafaie and Kober's large systematic review and meta-analysis linked laboratory drug-cue and craving responses with subsequent drug-use and relapse outcomes across studies, while documenting heterogeneity and the effects of publication-bias adjustments. Association with later behavior gives these measures genuine prospective relevance, but it does not establish that a particular cognitive bias causes relapse or that an individual person's outcome can be predicted accurately from a single laboratory measurement Vafaie & Kober (2022).

Null prospective results are equally informative. Snelleman and colleagues' study of alcohol dependence reported that measured attentional and approach tendencies did not predict relapse or time to relapse in its sample. Such findings are not an erasure of cue-related learning; they show that laboratory measures, exposure environments, treatment, and competing goals may occupy different levels of explanation. A productive science must treat prediction as an empirical outcome that requires validation rather than as a rhetorical consequence of an attractive mechanism Snelleman et al. (2015) Vafaie & Kober (2022).

Executive Function: Unity, Diversity, and Addiction Specificity

Executive function comprises partially separable processes supporting goal maintenance, updating, shifting, selection, and restraint. Laboratory measures are often treated as interchangeable proxies for a single self-control capacity, although their performance requirements differ. A stop-signal task concerns cancellation under a deadline, a go/no-go task measures omission under an asymmetric response schedule, and working-memory tests concern the maintenance and manipulation of information. The cognitive-science background matters because a difference in one paradigm need not generalize to another. Addiction-specific models must predict which component is affected, under which conditions, and why Smith et al. (2014) Verbruggen et al. (2019) Goldstein & Volkow (2011).

Research designs commonly compare a substance-use-disorder group with a nonclinical group. Such contrasts can identify a difference in average performance but generally cannot establish the temporal direction of causation. Exposure history, educational opportunity, socioeconomic adversity, psychiatric comorbidity, medication, nicotine co-use, and acute substance state can influence executive performance. Some such factors precede substance problems; others follow or accompany them. An account that attributes every observed difference to neurotoxicity or a fixed inhibitory deficit bypasses the required causal analysis Liu et al. (2019) Smith et al. (2014).

The distinction between general and substance-specific control is particularly important. A participant may perform adequately when the stop cue is a neutral symbol yet experience more interference when alcohol or cocaine cues are embedded in the response environment. Conversely, someone with generally slower motor responses might show impairment across both neutral and addiction-related tasks. To discriminate these alternatives, studies require cue-condition interactions, carefully matched comparison stimuli, within-person changes, and validation of the baseline control process Ceceli et al. (2026) Goldstein & Volkow (2011).

Go/No-Go Tasks: Withholding Prepotent Responses

In go/no-go tasks participants respond on frequent go trials and withhold responses on rarer no-go trials. Frequent preparation establishes a tendency to act, making the omission demand measurable through commission errors, response times, and sometimes neural signals. Addiction research uses both neutral and substance-related variants, but the construct is task dependent. Commission errors may reflect lapses of attention, incomplete understanding, risky speed emphasis, or failure to establish the required no-go mapping, as well as difficulty suppressing a response Smith et al. (2014) Verbruggen et al. (2019).

A group difference in no-go commission errors becomes more informative when go-trial performance and speed–accuracy tradeoffs are reported. Slower go responses can reduce errors without increasing inhibitory competence, while a participant responding quickly may appear disinhibited despite strategic speed emphasis. Investigators should compare error distributions across stimulus conditions, track practice effects, control visual features, and consider trial-by-trial adaptations. A single no-go accuracy statistic cannot support a general claim about a person's ability to resist substance use Hedge et al. (2018) Enkavi et al. (2019).

Drug-related no-go procedures can also create associations between a particular stimulus and inaction. That learning process matters for training studies, where repeated inhibition of one image may influence later responses to that image. It should be distinguished from an increase in a domain-general executive resource. Interventions that rely on such associations must demonstrate transfer to choices involving other stimuli and settings, and then to outcomes that matter clinically Iannazzo et al. (2025) Di Rosa et al. (2026).

Stop-Signal Tasks: Action Cancellation and Race Models

The stop-signal task introduces an occasional signal after the go stimulus. The delay between go and stop signals is varied to make cancellation harder or easier. The independent race model treats the completion of the go process and the stop process as competing events; a response occurs when going finishes first. The stop process's latency is not directly observed on successful stop trials, so stop-signal reaction time is an estimate derived from the observed go response-time distribution and the probability of successful stopping. The method's elegance rests on assumptions that must be tested rather than presumed Verbruggen et al. (2019).

Attention to the stop signal, proactive slowing, strategic waiting, and failure to initiate stopping can distort estimates. A participant may delay go responses because stopping is anticipated, producing a favorable apparent stopping score for reasons other than a faster cancellation mechanism. Likewise, ignoring some stop cues can be misread as uniformly slow inhibition. Contemporary consensus guidance emphasizes tracking response distributions, maintaining suitable stopping probabilities, diagnosing violations, and reporting analysis choices. These principles apply directly to addiction-group comparisons and treatment studies Verbruggen et al. (2019).

Meta-analytic evidence indicates differences in behavioral inhibition for some substance groups and some tasks, while the pattern is heterogeneous. Smith and colleagues reviewed a large corpus of inhibition studies, and subsequent pooled analyses demonstrated how substance classification, participant characteristics, and adjustment for confounders change interpretations. A general story of uniformly impaired inhibition across all forms of use therefore has insufficient precision. Carefully selected clinically characterized samples and explicit cue–control interactions represent more promising tests Smith et al. (2014) Liu et al. (2019).

Conflict Monitoring, Flexibility, and Rule Maintenance

Many everyday addiction-related situations require changing a plan when an expected reward becomes costly, suppressing a familiar response when it conflicts with current goals, or maintaining a longer-term intention while confronting a salient cue. Experimental tasks approximate components of these operations through switching costs, interference trials, error monitoring, and rule updates. The relevant constructs overlap but are not equivalent. A large switch cost may reflect processing of a new rule, changes in working-memory load, cautious response thresholds, or time required to disengage attention Goldstein & Volkow (2011) Ceceli et al. (2026).

Neuropsychological evidence invites a multidimensional account of control. A person may show impaired inhibition but intact switching, or intact neutral switching yet slower updating after a personally meaningful cue. Findings from gambling-disorder meta-analysis support impairment in some executive domains, including inhibition and shifting, without licensing the assertion that every domain or every patient is equally affected. Experimental design should include both converging tasks and differential predictions about mechanisms Peixoto et al. (2025).

Clinical translation must identify the situation in which control matters. Remembering a treatment goal, having access to a viable alternative, detecting a rising urge, and delaying an action can be separable intervention targets. Laboratory flexibility may contribute to these capacities but does not substitute for observing how behavior adapts under consequences and support. Ecologically valid measures combine the cognitive profile with opportunities, stressors, and repeated observations outside the laboratory Vafaie & Kober (2022) Schulte et al. (2014).

Working Memory, Prospection, and Competing Goals

Working memory enables the temporary maintenance and manipulation of information that guides behavior. Its relevance to addiction lies in the need to keep remote goals and anticipated consequences available while proximal cues activate familiar actions. However, low performance on a working-memory task may reflect attention lapses, fatigue, distress, withdrawal, or differences in general processing speed. The construct must be defined by particular task demands and interpreted with controls for baseline ability and clinical state Goldstein & Volkow (2011) Ceceli et al. (2026).

Future-oriented cognition is related but distinct. Remembering a rule, vividly imagining a future outcome, computing its subjective value, and implementing an intention at the right moment are different operations. A person may know the harmful consequences of a substance yet still allocate decision weight toward an immediate benefit because of withdrawal, immediate availability, social reinforcement, or altered uncertainty. Explanations framed exclusively as a lack of knowledge disregard this separation MacKillop et al. (2011) Bechara et al. (2001).

Studies would gain from measuring both the representational and implementation components of control. Trials could manipulate future-outcome reminders, alternative rewards, and cue exposure while collecting repeated measures of working memory, choice, and subsequent behavior. Such designs could determine whether a successful intervention makes future goals easier to retrieve, changes the values assigned to options, or modifies the policy translating a goal into action. Different mechanisms can yield identical improvements on a single laboratory task Pan et al. (2026) Iannazzo et al. (2025).

Impaired Response Inhibition and Salience Attribution

The impaired response inhibition and salience attribution framework links altered evaluation of drug cues to control processes involving prefrontal networks. Rather than treating salience and inhibition as independent defects, it proposes a context-dependent interaction in which addiction-related information receives disproportionate priority and the regulation of competing responses can be less effective. The model has motivated research on orbitofrontal, anterior cingulate, lateral prefrontal, and inferior frontal functions and their connections with other neural systems Goldstein & Volkow (2011) Ceceli et al. (2026).

The 2026 synthesis by Ceceli and colleagues emphasizes measuring interactions between cue context and cognition, individual variability, metacognition, recovery-related changes, and ecological validity. These priorities are empirically productive because they replace broad statements about a damaged control center with testable factorial predictions. For example, the hypothesis that a person's performance deteriorates selectively in drug-related trials is different from the hypothesis that performance is uniformly slower across all conditions Ceceli et al. (2026).

Neural activity is a constraint on mechanism rather than a diagnosis. Functional MRI signals are indirect hemodynamic measures; activation differences can track salience, effort, attention, prediction, motion, or compensatory performance. Neural group contrasts require carefully matched behaviors and control of substance state. The current evidence supports distributed interacting systems, alongside important uncertainty about individual biomarkers and causal directions. In particular, apparent recovery in some neural responses must be studied in relation to stable clinical outcomes Goldstein & Volkow (2011) Ceceli et al. (2026) Fascher et al. (2026).

Neural Systems, Causal Inference, and Reverse Inference

Prefrontal, striatal, insular, cingulate, and amygdala-linked circuits participate in attention, valuation, action preparation, emotional learning, and interoception. An addiction-related cue can recruit several of these operations concurrently. A regional hemodynamic difference cannot assign a unique mental process because the implicated neural tissue contributes to multiple tasks and contexts. Neural network models should therefore specify behaviorally distinguishable predictions, task contrasts, and conditions under which a signal is expected to reverse or disappear Goldstein & Volkow (2011) Ceceli et al. (2026).

Comparing people with and without a substance use disorder involves selection problems. Substance exposure may contribute to a measured difference, but preexisting vulnerabilities, cumulative stress, nicotine smoking, psychiatric comorbidity, and scanner-specific artifacts may also contribute. A 2026 meta-analysis of case-control drug-cue neuroimaging examined the convergence and limitations of contrasts across studies; its results reinforce caution about broad anatomical signatures inferred from heterogeneous case-control tasks Fascher et al. (2026).

Causal evidence becomes stronger when conditions are manipulated prospectively, when a change in neural processing precedes and mediates a behaviorally meaningful outcome, or when mechanistic predictions converge across independently designed measurements. Even so, mediation is not automatic proof of mechanism if the mediator is confounded or measured poorly. Circuit-level experiments in animals add important causal information, while human field validation remains essential for claims about treatment, voluntary action, and recovery Robinson & Berridge (2025) Everitt & Robbins (2016).

Evidence Heterogeneity Across Substances and Severity

Alcohol, nicotine, opioids, stimulants, cannabis, and sedative drugs differ in pharmacology, use patterns, acute effects, withdrawal physiology, and the populations typically enrolled in research. The construct of inhibitory control therefore cannot be assumed to display a constant effect size across these groups. Smith and colleagues' 2014 meta-analysis found a heterogeneous pattern of inhibition findings, while the 2019 mega-analysis by Liu and colleagues demonstrated the importance of controlling for participant-level confounders in substance-use comparisons Smith et al. (2014) Liu et al. (2019).

More recent targeted syntheses provide additional detail. A 2026 meta-analysis of cocaine use and cognitive inhibition examined studies using defined inhibition measures and reported a meaningful average difference, while still exposing the uncertainty generated by tasks, samples, severity, and comorbidity. The findings support a research hypothesis about cognitive inhibition in certain cocaine-use populations; they cannot be transferred without qualification to every substance, every level of consumption, or every stage of recovery Lopez-Martin et al. (2026).

Distinguishing hazardous use, a current diagnosed disorder, remission, treatment seeking, and long-term abstinence is methodologically indispensable. Participants who volunteer for treatment may differ systematically from people whose substance use is unrecognized or managed outside clinical services. Severity may correlate with task performance because of reciprocal influences among exposure, socioeconomic opportunity, mental health, and continued use. Multi-site sampling, transparent diagnostic characterization, state measures, and longitudinal within-person contrasts improve interpretive specificity Liu et al. (2019) Schulte et al. (2014).

Decision Processes: Choice, Expected Value, and Consequences

A decision process begins with the representation of available actions and continues through evaluation, selection, execution, and learning from outcomes. Addiction can alter these stages through multiple pathways: conditioned cues can draw attention to one option, withdrawal can increase the immediate value of relief, uncertainty can change risk assessment, and established routines can lower the effort required to enact a familiar choice. The same observed decision may therefore have different causes across individuals and occasions Everitt & Robbins (2016) Goldstein & Volkow (2011) MacKillop et al. (2011).

Models of economic choice often represent alternatives by subjective values that are dynamically influenced by delay, probability, effort, and state. Their usefulness lies in making assumptions explicit. A participant might choose a smaller immediate reward because of strong discounting of delay, a high valuation of the current option, uncertainty about the larger reward, or weak confidence in future access. Each possibility generates different predictions under manipulated delays and probabilities. Laboratory choices reveal local preferences and model parameters; their interpretation as enduring clinical traits requires independent validation MacKillop et al. (2011).

Decision-making research benefits from separating learning performance from choice performance. A person may know which option is advantageous but be prevented by contextual constraints from selecting it, or may choose efficiently given misleading beliefs about actual consequences. Multiple tasks, repeated observations, and explicit characterization of reward contingencies can illuminate these alternatives. The goal of explanation is a causal account of how information is converted into action, not a moral classification of a decision maker Bechara et al. (2001) Ceceli et al. (2026).

The Iowa Gambling Task and Somatic-Marker Interpretations

The Iowa Gambling Task was developed within research on decision-making after ventromedial prefrontal damage. Participants learn, through repeated selections, the consequences of choosing from decks that differ in apparent reward and long-run loss. The 1994 lesion study established an influential paradigm for examining decisions in which advantageous performance develops over trials. Later work compared individuals with alcohol and stimulant use histories against unaffected controls and patients with focal lesions, reporting group differences that helped motivate hypotheses about addiction-related future-consequence processing Bechara et al. (1994) Bechara et al. (2001).

Interpretation requires distinguishing performance from mechanism. Poor task outcomes can arise from differences in reversal learning, ambiguity tolerance, risk preference, reinforcement learning, memory, exploration strategies, or responsiveness to losses. It is tempting to infer that a person with substance problems possesses the same neural lesion as an injured patient, but a behavioral resemblance cannot establish anatomical equivalence. The causal evidence from lesions and the observational evidence from addiction groups occupy separate levels of inference Bechara et al. (1994) Bechara et al. (2001).

Longitudinal validation would ask whether task parameters precede future substance-related harm, predict changes after abstinence, and retain incremental value beyond established clinical indicators. Researchers should also measure trial trajectories because participants can discover contingencies at different rates. A summary score conceals whether impaired performance reflects initial exploration, a failure to reverse a choice after losses, or sustained preference for immediate payoff Bechara et al. (2001) MacKillop et al. (2011).

Intertemporal Choice and Delay Discounting

Intertemporal choice concerns tradeoffs between outcomes available at different times. Delay discounting describes how the subjective value of an outcome decreases as the wait for receipt increases, often approximated by a hyperbolic function. In substance-use research, larger discounting parameters have frequently been interpreted as stronger preference for immediate rewards. MacKillop and colleagues' 2011 meta-analysis synthesized associations between delay discounting and addictive behavior across a broad literature, supporting a meaningful but heterogeneous relationship MacKillop et al. (2011).

Discounting must be understood in the participant's environment. A delayed reward can be uncertain, resources may be scarce, and promises may have different credibility across settings. Preferences measured with hypothetical money can differ from those involving actual substances, social rewards, withdrawal relief, or recovery opportunities. Behavioral implications therefore depend on whether a task captures stable temporal preference, learned uncertainty, state-dependent valuation, or a combination. The task's experimental parameters and context should be reported alongside model estimates MacKillop et al. (2011).

Delay discounting intersects with executive control when future goals must remain accessible during choice, yet a steep discount curve is not equivalent to a failure to inhibit a motor response. Someone can cancel an instructed action efficiently while discounting delayed rewards heavily, and someone with strong long-run preferences may struggle with stop-signal detection. Conflating these constructs obscures testable individual differences and can misdirect interventions Verbruggen et al. (2019) Smith et al. (2014) MacKillop et al. (2011).

Habit Learning, Goal-Directed Control, and Compulsive Persistence

Habit formation and goal-directed action provide complementary explanatory frameworks for repeated substance behavior. In experimental work, goal-directed performance is sensitive to the current value of an outcome and the contingency linking an action to it; a habit can persist when a learned cue triggers responding despite changes in those contingencies. Everitt and Robbins' synthesis considers how repeated drug seeking can recruit habitual processes and, in some circumstances, become difficult to regulate despite adverse consequences. These mechanisms operate at the level of learned action policies and should be differentiated from subjective craving Everitt & Robbins (2016).

An action repeated many times is not necessarily habitual in the strict experimental sense, and clinically persistent use does not establish insensitivity to outcomes. People may continue use because the immediate relief is highly valued, because alternatives are constrained, because a social environment reinforces exposure, or because they expect long-term change to be difficult. To demonstrate habit, a study should test sensitivity to relevant devaluation or contingency manipulations rather than assume a habit from repetition or diagnostic status Everitt & Robbins (2016) Robinson & Berridge (2025).

Attention and executive control can influence the transition between strategies. A conditioned cue may rapidly retrieve a previously rewarded action; maintaining a changed goal can support alternative selection; withdrawal may change the relative value of avoiding discomfort. A useful integrated account predicts when behavior will remain flexible and when it will become less sensitive to changing outcomes. Clinical recovery, in turn, is compatible with learning new stable routines rather than requiring continuous effortful inhibition of every cue Everitt & Robbins (2016) Schulte et al. (2014).

State Dependence: Intoxication, Withdrawal, Stress, and Sleep

Performance on cognitive tasks depends on state. Acute intoxication can alter arousal, attention, psychomotor speed, judgment, and error monitoring, with effects that differ markedly among substances. Withdrawal may introduce physiological distress, negative affect, concentration problems, sleep disruption, and heightened motivation for relief. Stress can narrow attention toward immediate salient concerns and alter expectations about resources. A cross-sectional group difference recorded under these conditions cannot automatically be assigned to a durable executive deficit Goldstein & Volkow (2011) Ceceli et al. (2026).

Study design must specify recent exposure and abstinence windows, toxicology verification where ethically and scientifically appropriate, current medications, sleep, nicotine use, and relevant psychiatric symptoms. Uniform exclusion rules can improve certain comparisons but also select an atypical sample. The investigators' task is to identify the intended causal estimand: an effect of acute substance state, a characteristic of a clinically defined disorder, or an enduring difference associated with cumulative exposure Liu et al. (2019) Schulte et al. (2014).

Within-person designs can be especially informative when repeated observations are obtained during change. A person's performance may improve as acute effects resolve while other symptoms persist or shift. Longitudinal work requires control for practice effects, regression to the mean, selective attrition, and changes in environmental opportunities. Without those checks, apparent cognitive recovery can reflect improved familiarity with tasks or differences in who returns for assessment Schulte et al. (2014) Hedge et al. (2018).

Developmental Pathways and Preexisting Vulnerability

Adolescence and emerging adulthood involve changes in reward learning, social incentives, planning, and behavioral regulation. These changes can interact with substance availability, peer context, adversity, and individual learning histories. The dual-process tradition has emphasized automatic motivational responses alongside developing controlled processes, but those constructs cannot be assigned one-to-one to particular brain regions or used to predict inevitable disorder. Wiers and colleagues' developmental review proposed how automatic and controlled components might jointly shape risk trajectories Wiers et al. (2007).

Longitudinal research must distinguish predisposition from consequence. Lower inhibitory performance before heavy substance use might indicate vulnerability, but a post-exposure finding might instead reflect state effects, cumulative exposure, comorbidity, or selection. A third process—such as persistent adversity or another psychiatric condition—may contribute to both. Prospective cohort designs, sibling comparisons, repeated cognitive assessments, and accurate exposure records improve causal interpretation, although they cannot remove every unmeasured influence Wiers et al. (2007) Liu et al. (2019).

Developmental explanations should preserve the role of changing opportunities. Access to safer alternatives, social support, stable resources, and effective treatment can change behavior even when measured cognitive tendencies remain imperfect. A theory that takes the environment seriously predicts the conditions under which executive vulnerabilities become consequential, rather than treating a task-derived score as an individual's destiny Everitt & Robbins (2016) Ceceli et al. (2026).

Comorbidity, Polysubstance Exposure, and Selection Bias

Clinical substance-use samples often include people with histories of more than one substance, as well as co-occurring anxiety, mood, attention-related, or other mental-health difficulties. This makes strict attribution of cognitive differences to one drug difficult. Nicotine smoking, alcohol use, prescribed medications, sleep impairment, and stress can all covary with the disorder under investigation. Matching participants on a few demographic characteristics leaves many potential confounders unresolved Liu et al. (2019) Smith et al. (2014).

Controlling for confounders requires substantive causal reasoning. Some variables are prior common causes, some are consequences of prolonged use, and some are selection mechanisms created by treatment seeking. Adjusting for a mediator can obscure a real pathway; adjusting for a collider can introduce bias. Investigators should state whether they seek the total association, the causal effect of exposure, or an incremental prediction model. These are different research targets with different requirements Liu et al. (2019).

Polysubstance studies also challenge a binary case-control design. Exposure varies along age at initiation, intensity, route, recency, comorbidity, and treatment status; a single 'user' category can combine clinically distinct people. Multilevel modeling and transparent subgroup definitions can reveal whether deficits are shared, substance-specific, state-dependent, or chiefly associated with other conditions. The strength of the answer depends on sampling and data quality rather than the number of neuropsychological tests administered Smith et al. (2014) Lopez-Martin et al. (2026).

Gambling Disorder as a Cross-Domain Test

Gambling disorder provides a valuable test of addiction-related cognitive hypotheses because it involves persistent behavior and reward contingencies without the direct pharmacological effects of an ingested addictive substance. Researchers can examine cue salience, response inhibition, cognitive flexibility, reward learning, and risky decision-making while reducing some substance-intoxication confounds. However, financial, social, and motivational conditions differ across participants and gambling types, and the absence of direct drug exposure does not make the design causally simple Peixoto et al. (2025).

A 2025 meta-analysis by Peixoto and colleagues summarized neuropsychological findings in gambling disorder and identified difficulties in some executive domains, including inhibition and shifting. The synthesis supports cross-domain overlap while retaining heterogeneity and measurement limitations. The same task score need not reflect the same underlying mechanism in substance use and gambling: gambling outcomes are probabilistic, loss chasing and financial constraints alter valuation, and environmental access can be continuously mediated by online platforms Peixoto et al. (2025).

Cross-domain comparisons can clarify which cognitive processes are broadly related to addictive persistence and which depend on exposure, withdrawal, or particular reward structures. Strong designs should measure both general executive processes and domain-specific cues, establish matching on relevant confounders, and evaluate whether a proposed marker predicts impairment and change. A shared label should not substitute for evidence of shared mechanisms Ceceli et al. (2026) Peixoto et al. (2025).

Measurement Reliability, Construct Validity, and the Reliability Paradox

A reproducible average experimental effect does not automatically yield a reliable person-level measure. A task with a strong effect of stimulus condition can exhibit little stable variance between individuals, especially when performance approaches ceiling, trial counts are low, or noise dominates small differences. Hedge, Powell, and Sumner called attention to this reliability paradox, while Enkavi and colleagues' large-scale analysis demonstrated important limitations in test–retest reliability across commonly used measures of self-regulation Hedge et al. (2018) Enkavi et al. (2019).

In addiction research this distinction changes the interpretation of cognitive biomarkers. A task may consistently show that alcohol cues slow color naming in a sample while failing to rank individual patients consistently across repeated sessions. Such a measure may remain useful for testing a manipulation but have limited utility for screening, monitoring, or predicting an individual relapse. The intended application must determine the required reliability standard; one cannot borrow a group's statistically significant effect as proof of person-level precision Field et al. (2009) Hedge et al. (2018).

Validation requires sufficient trials, stable task instructions, transparent exclusion criteria, measurement invariance across groups, and independent replication. Researchers should report reliability and uncertainty for difference scores, not merely raw response times, because subtracting two noisy measurements may magnify error. Latent-variable models and multiple tasks can sometimes improve identification of common constructs, but their usefulness depends on adequate indicators, justified factor structures, and external outcome validation Enkavi et al. (2019) Verbruggen et al. (2019).

Longitudinal Prediction, Ecological Validity, and Relapse Outcomes

The clinical importance of an attentional or executive measure lies partly in whether it explains future behavior beyond information already available from treatment history, severity, craving, access, stress, and support. Prospective prediction needs a clearly defined observation period, an operational outcome, and appropriate handling of missing data. Relapse itself is variably defined as any use, return to hazardous use, recurrence of disorder, or treatment-relevant deterioration. Studies using these different outcomes cannot be treated as estimating a single effect Vafaie & Kober (2022) Snelleman et al. (2015).

Vafaie and Kober's 2022 meta-analysis found associations between laboratory cue responses and subsequent substance use or relapse, with evidence of heterogeneity and publication-bias concerns. In contrast, Snelleman and colleagues' prospective alcohol-dependence study found that their attentional and approach measures did not predict relapse timing. Taken together, these results favor an account in which some cue-related responses have population-level prospective information while predictive utility varies by instrument, cohort, context, and outcome definition Vafaie & Kober (2022) Snelleman et al. (2015).

Ecological momentary assessment and passive or participant-reported environmental sampling can provide repeated measures of craving, affect, cues, exposure opportunity, and subsequent action. Their strengths include temporal detail and within-person analysis. Their challenges include missingness at high-risk moments, privacy, recall even during short prompts, and reactivity to repeated measurement. Combining laboratory tasks with repeated real-world observations allows researchers to ask whether a marker generalizes to settings in which addiction-related choices actually occur Ceceli et al. (2026) Vafaie & Kober (2022).

Recovery, Abstinence, and Change in Cognitive Performance

Recovery involves changes in exposure, physiological state, environmental circumstances, learned expectations, coping, and social support. Cognitive performance may change as acute intoxication or withdrawal resolves, as sleep and stability improve, and as alternative routines become established. Schulte and colleagues' review of prospective neurocognitive studies after sustained abstinence described partial recovery in several functions while emphasizing variability and limitations in longitudinal evidence. It does not support a universal timeline or a guarantee of complete normalization for every substance and individual Schulte et al. (2014).

Repeated testing can produce improvement through familiarity, learned strategy, or reduced anxiety, independently of a change in the underlying process. Appropriate control groups, alternate task forms, and modeling of practice effects help distinguish these pathways. The absence of an observed deficit during abstinence may indicate recovery, selection of those who achieved abstinence, or insensitivity of the task. Interpretation depends on where an individual's trajectory begins and which participants remain under observation Schulte et al. (2014) Hedge et al. (2018).

The practical implication is to treat cognitive control as dynamic and contextual. High cue reactivity at one point does not establish a permanent inability to regulate action, and a favorable test at one follow-up does not settle the question of continuing risk. Mechanistic models should explain repeated transitions among stability, changing motivation, exposure, treatment engagement, and return to use. The appropriate scientific unit is often the person-in-context over time Ceceli et al. (2026) Vafaie & Kober (2022).

Cognitive Bias Modification: From Experimental Manipulation to Clinical Outcome

Cognitive bias modification attempts to retrain responses to addiction-related information. Attention training can systematically redirect the location of a probe away from substance cues; approach–avoidance training can associate drug cues with pushing rather than pulling; response-inhibition training repeatedly maps selected stimuli to withholding. The primary experimental outcome is whether the trained bias changes. The clinically meaningful outcome is whether behavior, functioning, sustained recovery, or a clearly defined relapse measure improves beyond a credible comparison condition Wiers et al. (2011) Pan et al. (2026).

Wiers and colleagues' randomized study reported that retraining alcohol-related action tendencies in an inpatient context altered the targeted bias and was associated with improved one-year outcomes. Subsequent research expanded delivery formats, populations, cue classes, and comparison conditions, producing results that vary by intensity and clinical setting. The 2026 individual-participant-data Bayesian meta-analysis by Pan and colleagues synthesized 23 studies involving 8,297 participants and found evidence for changes in bias and relapse risk, while moderator adjustment weakened the overall conclusion and clear reductions in drinking were not established Wiers et al. (2011) Pan et al. (2026).

These results warrant a bounded interpretation: training can change certain learned laboratory responses under some circumstances, and carefully specified clinical contexts may derive benefit, but a broad claim that training alone treats addiction would exceed the evidence. Active comparators, participant expectations, treatment co-interventions, missing outcomes, and selective reporting deserve particular scrutiny. Demonstrating a change in an intermediate mechanism is valuable when it is accompanied by causal mediation evidence and independent replication in outcomes that matter Pan et al. (2026) Vafaie & Kober (2022).

Inhibitory-Control Training and Transfer Beyond the Task

Inhibitory-control training asks whether repeated response withholding or cancellation can reduce unwanted behavior. Different procedures target general stopping capacity, automatic stimulus–stop associations, or strategic caution. Improvements in trained accuracy can reflect learning of the response mapping without a change in general inhibitory ability. The appropriate mechanism test therefore compares performance on untrained cues and tasks and, ultimately, outcomes in the relevant clinical population Iannazzo et al. (2025) Verbruggen et al. (2019).

A 2025 meta-analytic review by Iannazzo and colleagues considered inhibitory-control training across appetitive behaviors and identified generally small, conditional behavioral effects. A 2026 meta-analysis by Di Rosa and colleagues focused on clinical and at-risk populations and did not find clear evidence that stand-alone inhibitory training produced meaningful reductions in unhealthy behavior across the reviewed studies. These conclusions differ in sample frame and outcome, and together they emphasize the limits of assuming clinical transfer from training-task success Iannazzo et al. (2025) Di Rosa et al. (2026).

An intervention study must specify how transfer is hypothesized to occur. If improvement rests on direct stimulus–stop associations, the effects may remain tied to the trained images or situations. If improvement rests on improved general stopping, the researcher needs tasks and psychometric evidence supporting cross-context change. If a clinical benefit occurs through altered expectations or engagement with concurrent treatment, those pathways should be evaluated rather than credited automatically to strengthened inhibition Iannazzo et al. (2025) Di Rosa et al. (2026).

Clinical Integration: Assessment, Treatment, and Ethical Interpretation

Cognitive findings can enrich the scientific understanding of why particular situations are difficult, but they currently provide limited grounds for independent individual diagnosis or treatment selection. Clinical decision-making must integrate the diagnosed condition, substance, current safety, co-occurring disorders, treatment history, motivation, resources, and context. The relevant clinical question may concern how to anticipate a cue-rich environment, improve access to alternative reinforcement, or support planned behavior—not whether a person has passed an isolated laboratory inhibition test Goldstein & Volkow (2011) Ceceli et al. (2026).

Intervention effectiveness must be established through condition-specific clinical research. Cognitive-bias and inhibition training are experimental or adjunctive approaches with heterogeneous results; they cannot be assumed to replace interventions with established disorder-specific benefits. In particular, acute intoxication, potentially dangerous withdrawal, and overdose risk require appropriate medical assessment. A review of cognitive processes supplies a conceptual basis for research and care planning, not individualized prescribing or a substitute for emergency or specialist evaluation Pan et al. (2026) Di Rosa et al. (2026).

Ethical reporting avoids language that equates cognitive differences with moral weakness, absence of agency, or irreversibility. Research participants may perform differently because of social adversity, medication, fatigue, or current circumstances, and even a robust group difference does not describe every member of the group. Any practical use of predictive technology requires validated accuracy, attention to unequal error rates, privacy safeguards, clear human responsibility, and the possibility of contesting consequential assessments Hedge et al. (2018) Enkavi et al. (2019) Ceceli et al. (2026).

Computational Models and the Distinction Between Parameters and Causes

Computational modeling formalizes hypotheses about how information yields action. Reinforcement-learning models can estimate how outcomes update expected values; evidence-accumulation models can describe decisions under variable information; race models represent stopping under time pressure; temporal-discounting models specify how delay changes value. A fitted parameter is a useful summary of observed decisions under the chosen mathematical assumptions. It is not automatically a neural mechanism, a stable personality attribute, or a clinical cause Verbruggen et al. (2019) MacKillop et al. (2011).

Model comparison should use predictions on held-out trials or external datasets when feasible, recoverability tests, uncertainty estimates, and sensitivity analyses. Competing models may fit the same observed choices, especially when tasks contain few trials or limited variation in rewards and delays. Combining eye movement, response times, choices, and prospective observations can help discriminate explanations, but increased measurement complexity introduces its own sources of missingness and overfitting Ceceli et al. (2026) Hedge et al. (2018).

A promising research program connects latent parameters to experimentally manipulable contingencies. If a model attributes persistent responding to failure to update an outcome's value, a controlled value-change manipulation should produce a predictable behavioral signature. If it attributes persistence to attentional priority, changing cue accessibility or task demand should modify the response. Such tests make theoretical interpretations more precise and protect against explaining every observed outcome after the fact Everitt & Robbins (2016) Robinson & Berridge (2025).

Artificial Era: Human–Artificial Decision Support and Cognitive Offloading

Digital systems increasingly structure which cues are encountered, when choices are offered, and how immediate feedback arrives. Interfaces can amplify a cue by repetition or personalization, shape attention through notifications, and modify action costs through frictionless access. The research relevance is the ability of an environment to change cognitive demands and learned response probabilities. The effects of a particular platform require empirical demonstration rather than inference from the general existence of personalization or automation Field & Cox (2008) Everitt & Robbins (2016).

Human–artificial decision support may also externalize elements of prospective planning: reminders, structured coping intentions, access to alternative information, and timely self-monitoring. Whether such arrangements improve recovery depends on usability, consent, data governance, the reliability of inference, and clinical outcomes. A system that detects many possible cues can still generate inaccurate individual predictions when the underlying cognitive markers have low retest reliability or poor external validity Hedge et al. (2018) Enkavi et al. (2019) Vafaie & Kober (2022).

The Artificial Era enlarges the scientific unit of analysis from an isolated individual and laboratory stimulus to coupled human, technological, and social decision environments. This is a research perspective requiring operational definitions: what information the system observes, which intervention it performs, how participants retain control, and what outcome changes compared with an appropriate alternative. Behavioral data should be treated as sensitive personal information, and AI-generated explanations should remain auditable hypotheses rather than unvalidated psychiatric determinations Ceceli et al. (2026) Pan et al. (2026).

Critical Scientific Synthesis and Research Priorities

The strongest general finding is that attention, valuation, learning, and executive operations interact in addiction-relevant contexts, with pronounced heterogeneity across substances, severity, task designs, and states. Drug cues can affect attention, some clinical samples show poorer inhibition or decision performance, and some cue measures carry prospective information. At the same time, modest craving–bias correlations, null individual prediction studies, weak reliability for some task contrasts, and attenuated clinical effects after bias-training adjustments constrain broad causal claims Field et al. (2009) Smith et al. (2014) Snelleman et al. (2015) Pan et al. (2026).

Four methodological priorities follow. First, studies should define the clinical outcome and causal estimand before interpreting a laboratory contrast. Second, they should measure both general executive performance and interactions with substance-associated cues, while accounting for intoxication, withdrawal, medication, comorbidity, and environment. Third, they should establish reliability, measurement invariance, and prediction on independent samples. Fourth, interventions should use active controls and demonstrate sustained improvement in meaningful outcomes rather than training-task gains alone Verbruggen et al. (2019) Enkavi et al. (2019) Di Rosa et al. (2026).

Mechanistic integration requires comparing explanations rather than aggregating them rhetorically. Incentive sensitization predicts changes in motivational attraction; habit accounts predict distinctive outcome-insensitivity under suitable conditions; executive accounts predict defined task and cue interactions; valuation models predict choice under changing contingencies. Carefully designed experiments may reveal that several pathways converge on a behavior or that their contributions differ by person and time. The value of the integrated science is a set of discriminating tests capable of identifying where and how regulation changes Robinson & Berridge (2025) Everitt & Robbins (2016) Ceceli et al. (2026).

Conclusion

Addiction-related attention, executive control, and decision processes form an interconnected and empirically investigable system through which learned cues, current states, competing goals, expected outcomes, and available actions influence behavior. Attention-bias tasks reveal interference or orientation under controlled conditions; inhibition tasks measure withholding or cancellation within formal assumptions; choice paradigms capture preferences and learning under specified incentives. Together they offer a substantial scientific vocabulary, provided that each observation is interpreted at its proper level Field & Cox (2008) Verbruggen et al. (2019) MacKillop et al. (2011).

The evidence favors context-sensitive explanations with explicit attention to individual variability and measurement limitations. It establishes neither a universal executive deficit nor a universally effective cognitive training remedy. The most convincing research connects laboratory performance with reliable measurement, longitudinal outcomes, mechanistic manipulation, and clinically meaningful change. Continued progress will come from distinguishing what causes action, what merely accompanies it, and what can be changed safely and durably through validated intervention Liu et al. (2019) Vafaie & Kober (2022) Pan et al. (2026) Di Rosa et al. (2026).

The scientific horizon extends to the Artificial Era, where technologies and environments participate in the distribution of attention and the organization of decisions. Their psychological consequences must be studied with the same rigor as any other intervention: precise constructs, transparent designs, replicated results, meaningful outcomes, and accountable use of evidence. The central achievement of this dossier is an integrated framework for evaluating cognitive addiction research while retaining the distinction between scientific explanation, empirical prediction, and responsible clinical practice Ceceli et al. (2026) Enkavi et al. (2019).

References

  • Bechara, A., Damasio, A. R., Damasio, H., & Anderson, S. W. (1994). Insensitivity to future consequences following damage to human prefrontal cortex. Cognition, 50(1–3), 7–15. https://doi.org/10.1016/0010-0277(94)90018-3
  • Bechara, A., Dolan, S., Denburg, N., Hindes, A., Anderson, S. W., & Nathan, P. E. (2001). Decision-making deficits, linked to a dysfunctional ventromedial prefrontal cortex, revealed in alcohol and stimulant abusers. Neuropsychologia, 39(4), 376–389. https://doi.org/10.1016/S0028-3932(00)00136-6
  • Berridge, K. C., & Robinson, T. E. (2016). Liking, wanting, and the incentive-sensitization theory of addiction. American Psychologist, 71(8), 670–679. https://doi.org/10.1037/amp0000059
  • Ceceli, A. O., Huang, Y., Kronberg, G., McClain, N., King, S. G., Butelman, E. R., Alia-Klein, N., & Goldstein, R. Z. (2026). The impaired response inhibition and salience attribution model of drug addiction: Recent neuroimaging evidence and future directions. Annual Review of Psychology, 77, 81–108. https://doi.org/10.1146/annurev-psych-040725-025923
  • Di Rosa, E., Ronconi, L., Del Carlo, B., Giordano, G., Macchia, E., Cardi, V., Gentili, C., Masina, F., Guarrera, C., Mapelli, D., & Edelstyn, N. (2026). Inhibitory control training and unhealthy behaviours: A meta-analysis testing short and long-term effects in clinical and at-risk populations. Scientific Reports, 16, Article 13928. https://doi.org/10.1038/s41598-026-43063-0
  • Enkavi, A. Z., Eisenberg, I. W., Bissett, P. G., Mazza, G. L., MacKinnon, D. P., Marsch, L. A., & Poldrack, R. A. (2019). Large-scale analysis of test–retest reliabilities of self-regulation measures. Proceedings of the National Academy of Sciences, 116(12), 5472–5477. https://doi.org/10.1073/pnas.1818430116
  • Everitt, B. J., & Robbins, T. W. (2016). Drug addiction: Updating actions to habits to compulsions ten years on. Annual Review of Psychology, 67, 23–50. https://doi.org/10.1146/annurev-psych-122414-033457
  • Fascher, M., Krasowski, K., & Muehlhan, M. (2026). What can case-control comparisons of neural drug cue reactivity tell us about addiction? A neuroimaging meta-analysis. Translational Psychiatry, 16, Article 501. https://doi.org/10.1038/s41398-026-04458-9
  • Field, M., & Cox, W. M. (2008). Attentional bias in addictive behaviors: A review of its development, causes, and consequences. Drug and Alcohol Dependence, 97(1–2), 1–20. https://doi.org/10.1016/j.drugalcdep.2008.03.030
  • Field, M., Munafò, M. R., & Franken, I. H. A. (2009). A meta-analytic investigation of the relationship between attentional bias and subjective craving in substance abuse. Psychological Bulletin, 135(4), 589–607. https://doi.org/10.1037/a0015843
  • Goldstein, R. Z., & Volkow, N. D. (2011). Dysfunction of the prefrontal cortex in addiction: Neuroimaging findings and clinical implications. Nature Reviews Neuroscience, 12(11), 652–669. https://doi.org/10.1038/nrn3119
  • Hedge, C., Powell, G., & Sumner, P. (2018). The reliability paradox: Why robust cognitive tasks do not produce reliable individual differences. Behavior Research Methods, 50, 1166–1186. https://doi.org/10.3758/s13428-017-0935-1
  • Iannazzo, L. H., Hayden, M. J., Lawrence, N. S., Kakoschke, N., Hughes, L. K., Van Egmond, K., Lum, J., & Staiger, P. K. (2025). Inhibitory control training to reduce appetitive behaviour: A meta-analytic investigation of effectiveness, potential moderators, and underlying mechanisms of change. Health Psychology Review, 19(1), 66–96. https://doi.org/10.1080/17437199.2024.2410018
  • Liu, Y., van den Wildenberg, W. P. M., de Graaf, Y., Ames, S. L., Baldacchino, A., Bø, R., Cadaveira, F., Campanella, S., Christiansen, P., Claus, E. D., Colzato, L. S., Filbey, F. M., Foxe, J. J., Garavan, H., Hendershot, C. S., Hester, R., Jester, J. M., Karoly, H. C., Kräplin, A., … Wiers, R. W. (2019). Is (poly-) substance use associated with impaired inhibitory control? A mega-analysis controlling for confounders. Neuroscience & Biobehavioral Reviews, 105, 288–304. https://doi.org/10.1016/j.neubiorev.2019.07.006
  • Lopez-Martin, G., González-Roz, A., Belisario, K., Gervilla, E., & Andrés, A. (2026). Systematic review and meta-analysis of the impact of cocaine use on cognitive inhibition. Neuropsychology Review, 36(3), 428–445. https://doi.org/10.1007/s11065-025-09685-3
  • MacKillop, J., Amlung, M. T., Few, L. R., Ray, L. A., Sweet, L. H., & Munafò, M. R. (2011). Delayed reward discounting and addictive behavior: A meta-analysis. Psychopharmacology, 216(3), 305–321. https://doi.org/10.1007/s00213-011-2229-0
  • Pan, T., Zhao, X., Bartoš, F., Larsen, H., Manning, V., Boffo, M., & Wiers, R. W. (2026). Cognitive bias modification in alcohol use disorder and problematic drinking: A revised and updated IPD Bayesian meta-analysis. Clinical Psychology Review, 124, Article 102709. https://doi.org/10.1016/j.cpr.2026.102709
  • Peixoto, M., Dores, A., Monteiro, M., Marques, A., & Barbosa, F. (2025). Executive function in gambling disorder: A meta-analysis on neuropsychological evidence. Journal of Gambling Studies, 41(2), 449–488. https://doi.org/10.1007/s10899-025-10383-1
  • Robinson, T. E., & Berridge, K. C. (1993). The neural basis of drug craving: An incentive-sensitization theory of addiction. Brain Research Reviews, 18(3), 247–291. https://doi.org/10.1016/0165-0173(93)90013-P
  • Robinson, T. E., & Berridge, K. C. (2025). The incentive-sensitization theory of addiction 30 years on. Annual Review of Psychology, 76, 29–58. https://doi.org/10.1146/annurev-psych-011624-024031
  • Schulte, M. H. J., Cousijn, J., den Uyl, T. E., Goudriaan, A. E., van den Brink, W., Veltman, D. J., Schilt, T., & Wiers, R. W. (2014). Recovery of neurocognitive functions following sustained abstinence after substance dependence and implications for treatment. Clinical Psychology Review, 34(7), 531–550. https://doi.org/10.1016/j.cpr.2014.08.002
  • Smith, J. L., Mattick, R. P., Jamadar, S. D., & Iredale, J. M. (2014). Deficits in behavioural inhibition in substance abuse and addiction: A meta-analysis. Drug and Alcohol Dependence, 145, 1–33. https://doi.org/10.1016/j.drugalcdep.2014.08.009
  • Snelleman, M., Schoenmakers, T. M., & van de Mheen, D. (2015). Attentional bias and approach/avoidance tendencies do not predict relapse or time to relapse in alcohol dependency. Alcoholism: Clinical and Experimental Research, 39(9), 1734–1739. https://doi.org/10.1111/acer.12817
  • Vafaie, N., & Kober, H. (2022). Association of drug cues and craving with drug use and relapse: A systematic review and meta-analysis. JAMA Psychiatry, 79(7), 641–650. https://doi.org/10.1001/jamapsychiatry.2022.1240
  • Verbruggen, F., Aron, A. R., Band, G. P. H., Beste, C., Bissett, P. G., Brockett, A. T., Brown, J. W., Chamberlain, S. R., Chambers, C. D., Colonius, H., Cools, R., Corneil, B. D., Coxon, J. P., Dupuis, A., Eagle, D. M., Garavan, H., Greenhouse, I., Heathcote, A., Huster, R. J., … Boehler, C. N. (2019). A consensus guide to capturing the ability to inhibit actions and impulsive behaviors in the stop-signal task. eLife, 8, e46323. https://doi.org/10.7554/eLife.46323
  • Wiers, R. W., Bartholow, B. D., van den Wildenberg, E., Thush, C., Engels, R. C. M. E., Sher, K. J., Grenard, J., Ames, S. L., & Stacy, A. W. (2007). Automatic and controlled processes and the development of addictive behaviors in adolescents: A review and a model. Pharmacology Biochemistry and Behavior, 86(2), 263–283. https://doi.org/10.1016/j.pbb.2006.09.021
  • Wiers, R. W., Eberl, C., Rinck, M., Becker, E. S., & Lindenmeyer, J. (2011). Retraining automatic action tendencies changes alcoholic patients' approach bias for alcohol and improves treatment outcome. Psychological Science, 22(4), 490–497. https://doi.org/10.1177/0956797611400615