Key Takeaways
Treat high-THC cannabis as a dose-sensitive psychosis risk, especially for adolescents
The central claim. Berenson argues that cannabis can contribute to psychosis, with risk rising as use becomes heavier and begins earlier. In a Swedish study of 45,570 conscripts, 2.8 percent of those reporting at least 50 uses later developed schizophrenia, roughly six times the rate among nonusers. In the Dunedin study, cannabis users at age 15 were more than four times as likely to develop a psychotic disorder, and the association remained after researchers accounted for symptoms recorded at age 11.
Risk is not destiny. The National Academy of Medicine’s 2017 review judged the evidence for a cannabis-psychosis link strong and dose-related. The book stresses that most users will not develop schizophrenia, and that researchers cannot reliably predict which individuals are most vulnerable.
What's persuasive here is the convergence of dose patterns, timing, and repeated findings across different populations. Each makes a simple coincidence explanation less satisfying. But a population-level risk is not a forecast for a particular person, and these studies do not establish that cannabis is the sole cause in any individual case. Good risk communication needs both truths: a real increase in risk can matter greatly across millions of people, while most users never experience the outcome that researchers are measuring.
Do not equate today's high-potency products with 1970s marijuana
The product changed. Berenson contrasts the marijuana common in the 1970s, often below 2 percent THC, with dispensary flower that he says routinely reached 20 to 25 percent by the late 2010s. THC is the intoxicating compound in cannabis. Concentrated products such as wax and shatter can contain far higher levels. A familiar memory of sharing a weak joint, he argues, is a poor guide to the effects of these products.
Use changed, too. The book reports that about one in five cannabis users used daily or almost daily, compared with about one in fifteen drinkers. Daily or near-daily cannabis use grew from roughly three million Americans in 2005 to more than eight million in 2017. Potency and frequency together make exposure, not merely the word cannabis, central to assessing risk.
This is a useful correction to reasoning from personal experience. A person’s recollection of an older, weaker product cannot settle the safety of a substantially more concentrated one. The same principle applies to many drugs: dose, route, and frequency shape effects. Still, potency alone does not tell the whole story. A careful policy would also track how much people consume, how products are labeled, and whether users understand that edibles and extracts can produce effects unlike familiar smoked cannabis.
Trust converging evidence more than any single study or vivid case
Evidence accumulated in stages. Berenson begins with nineteenth-century reports from India and Mexico describing cannabis-related madness and violence. These accounts establish historical concern, not modern proof. Later case reports showed that cannabis could trigger temporary psychosis, but could not determine how often it happened or whether the drug caused lasting illness.
Better designs followed. The Swedish conscript research tracked tens of thousands of men over time. The Dunedin study repeatedly assessed a New Zealand birth cohort, including psychotic symptoms before many participants used cannabis. Dutch and other European studies added further populations and methods. Their shared findings strengthened the argument beyond any one dataset. The 2017 National Academy review synthesized thousands of studies and concluded that cannabis use likely increases psychosis risk, with higher use associated with greater risk.
The strongest reasoning is cumulative rather than anecdotal. Historical testimony can generate a question, and longitudinal research can test whether exposure precedes an outcome, whether risk changes with dose, and whether the pattern survives adjustment for other factors. That is a more credible route to causal judgment than either a dramatic case or a reassuring personal story. The remaining need is transparent replication and better measurement, especially in the United States, where the book describes major gaps in tracking psychotic disorders.
Medical branding can spread faster than evidence of medical benefit
A powerful reframing. California voters approved Proposition 215 in 1996, allowing medical cannabis use with a physician’s authorization. Berenson argues that this compassionate framing helped normalize cannabis and created a bridge to recreational legalization. In practice, he says, many authorizations were issued for loosely defined conditions, and a small number of cannabis-friendly doctors handled large shares of patients.
Medicine requires evidence. The book distinguishes THC, which produces intoxication, from CBD, a non-intoxicating cannabis compound. It cites the National Academy review as finding evidence for only a narrow set of benefits, not the broad claims that cannabis treats cancer, dementia, or many other conditions. A plant’s reputation as medicine, Berenson argues, should not substitute for controlled trials of specific compounds, doses, and uses.
The distinction between a treatment claim and a political claim matters. Compassion for patients can motivate research without proving that a particular product works, and an effective compound does not validate every product made from the same plant. This is a general lesson in evidence-based medicine: test the actual intervention, for a defined condition, at a known dose. The book’s critique is strongest when it asks for consistent standards. Its discussion is less complete when it treats weak evidence for some uses as evidence against every possible therapeutic use.
Separate decriminalizing possession from selling cannabis as a legal industry
Two different policies. Decriminalization reduces or removes criminal penalties for possessing small amounts, often replacing arrest with a ticket. Legalization permits a regulated commercial market where businesses can grow, advertise, and sell cannabis. Berenson argues that advocates often present these as one choice, even though someone can oppose harsh possession penalties while also questioning widespread retail access.
The political route mattered. The book describes how medical-marijuana campaigns attracted broad support, then helped build dispensaries, growers, and investors with an interest in full legalization. It also acknowledges the racial disparity in marijuana arrests that made reform urgent. Berenson’s proposed distinction is that unfair punishment can be addressed without treating a commercial market as the only alternative to prohibition.
This separation improves the debate by breaking a false binary. Criminal penalties can be reduced without automatically endorsing unrestricted sales, just as legal retail can be regulated without requiring imprisonment for possession. The challenge is designing the middle ground: enforcement discretion and unequal policing can persist even when penalties are lower. A useful policy discussion therefore asks separate questions about possession, sales, advertising, age limits, product strength, and tax incentives, rather than asking only whether cannabis should be legal or illegal.
Investigate cannabis-related violence without treating every user as dangerous
A proposed pathway. Berenson connects cannabis to violence through intoxication, paranoia, and psychosis. He cites studies associating cannabis use with aggression and research finding that psychosis is linked to higher rates of violent crime, especially when substance abuse is also present. In one Swiss study of young people with psychosis, nearly half of cannabis-dependent participants became violent over three years, compared with about 15 percent of those without cannabis dependence.
Cases illustrate, not prove. The book recounts Richard Kirk, who ate a high-dose THC edible and killed his wife during a psychotic episode. Kirk also had a history of opioid misuse, prior paranoia, a family history of schizophrenia, and access to a gun. Berenson presents the case as a warning, but those competing factors make it impossible to treat one tragedy as proof of a single cause.
The section usefully challenges the assumption that a drug associated with relaxation cannot also be associated with violence. Effects vary across people and situations, and laboratory studies of calm volunteers may not capture high-risk patients or real-world settings. Yet violence findings are often observational, and association does not establish that cannabis caused a particular assault or homicide. The ethical balance is to investigate risks carefully while refusing to turn a statistical association into suspicion of every person who uses cannabis.
Avoid THC as self-treatment during adolescence or psychiatric vulnerability
The highest-stakes caution. Berenson emphasizes adolescence because cannabis use often begins years before the age when psychotic disorders first become apparent. He argues that frequent use during this period may increase risk, while a family history of psychosis or previous paranoia can make caution especially important. The book also describes people continuing to use cannabis after symptoms begin, believing it relieves anxiety or distress even as clinicians observe worsening symptoms.
No reliable personal test. Doctors cannot currently identify every person who will develop psychosis after using cannabis. Berenson’s practical recommendation is to avoid use in young people and in those with psychiatric vulnerability, rather than assuming that a pleasant early experience predicts future safety. He also warns against treating THC as an antidepressant or anxiety remedy without strong clinical evidence.
This is a precautionary recommendation under uncertainty, not a claim that every adolescent user will become ill. The potential severity of psychosis makes age and vulnerability relevant in the same way that known risk factors guide other health decisions. But screening is imperfect, and family history does not determine an individual’s outcome. Clear warnings are more useful than moral judgment: they help young people and families weigh a preventable exposure without implying that people who develop psychosis are responsible for their illness.
Do not mistake state-level opioid correlations for proof cannabis prevents addiction
A popular claim weakened. A 2014 study reported that states with medical-marijuana laws had fewer opioid overdose deaths through 2010. Berenson argues that the result reflected geography: early medical-cannabis states were concentrated in the West, while the opioid crisis initially hit Appalachia and nearby regions. When researchers included later years, he says, the apparent protective association disappeared.
Better questions use people over time. The book cites a national survey that followed 34,000 adults and found cannabis users were more likely to begin using opioids over the next three years, even after adjustment for other factors. Heavy cannabis users had a higher reported risk still. These findings do not prove cannabis causes opioid addiction, but they undermine the confident claim that cannabis is an established substitute or solution.
This is a textbook warning about ecological inference: a pattern across states can arise from geography, timing, or other differences, and may not describe what happens to individual users. The book is right to demand scrutiny of a headline-grabbing policy claim. The same caution applies to its own state comparisons and cross-national contrasts. Neither a suggestive correlation nor its disappearance settles causality. Stronger evidence requires careful longitudinal research and, where ethical and feasible, well-designed trials of specific medical uses.
Make warnings and surveillance part of any cannabis policy
The data system is weak. Berenson describes the United States as lacking a reliable national count of schizophrenia and other psychotic disorders. He analyzes emergency-room records showing that cases with both psychosis and a cannabis-use diagnosis rose from about 30,000 in 2006 to nearly 90,000 in 2014. That increase is a signal worth investigating, not proof that cannabis caused every case.
Public health needs visibility. The book calls for independent research, clearer warnings, and more skeptical reporting about both benefits and harms. It also argues that young people should receive direct risk information rather than relying on slogans from either prohibitionists or industry advocates. Because the available data are incomplete, monitoring use patterns, product potency, emergency visits, and mental-health outcomes would help policymakers notice harms sooner and distinguish them from coincidence.
Surveillance is a practical response to uncertainty. A system that tracks both intended benefits and unintended harms can correct policy without waiting for a crisis to become undeniable. This is especially important when products and patterns of use change quickly. The book’s own emergency-room analysis is a useful example of asking a neglected question, but administrative diagnoses are imperfect proxies and cannot establish causation alone. Independent researchers should be able to test the signal, publish null findings, and make their methods available for scrutiny.
Analysis
Tell Your Children is a science-and-policy argument built around a narrative of belated recognition. It opens with an extreme case, travels through historical records and epidemiological studies, then returns to contemporary violence and legalization. Its central claim is that cannabis, particularly frequent use of high-THC products, can contribute to psychosis and may raise violence risk through psychosis, intoxication, and related pathways. Its most useful contribution is to force distinctions often blurred in public debate: old low-potency cannabis versus modern concentrates, THC versus CBD, medical research versus medical branding, and decriminalizing possession versus creating a retail industry.
The book’s strongest evidential thread is the psychosis argument. The Swedish conscript cohort, the Dunedin study, later European research, and the National Academy review provide evidence across methods and populations. The convergence matters more than any single case or statistic. The account also identifies an important public-health problem: weak national monitoring makes it difficult to determine whether changes in use correspond to changes in severe mental illness.
Its least secure move is from association to broad causal and policy conclusions. Longitudinal studies and dose-response patterns strengthen a causal interpretation, but residual confounding remains possible. The argument about violence relies heavily on observational studies, clinical samples, and selected criminal cases. These can identify serious signals, but cannot establish that cannabis caused each crime or quantify its contribution to national violence. State crime comparisons and opioid-death correlations are particularly vulnerable to geographic and temporal confounding. The book sometimes applies skepticism unevenly, demanding rigorous proof from legalization advocates while drawing forceful conclusions from suggestive data of its own.
That tension does not make the evidence irrelevant. It means the right response is neither dismissal nor certainty. The strongest reading is that cannabis risk deserves explicit warnings, rigorous independent study, and better surveillance, especially for adolescents and heavy users. The book is less conclusive as a case against every form of legalization. Readers can accept the need for honest risk communication while separately debating criminal penalties, commercial access, and regulation.
Review Summary
Tell Your Children presents a controversial perspective on marijuana, linking its use to psychosis and violence. Reviews are polarized, with supporters praising the book's research and warnings about marijuana's dangers, while critics argue it relies on correlation rather than causation and lacks proper sourcing. Many readers found the content eye-opening and important, while others dismissed it as fear-mongering. The book sparked debate about marijuana legalization and its potential risks, especially for young users and those predisposed to mental illness.
People Also Read
Glossary
Psychosis
break with shared realityA mental state in which a person has difficulty distinguishing internally generated experiences from shared external reality. It can involve hallucinations, such as hearing voices, or delusions, such as strongly held false beliefs. Psychosis can occur in several disorders and can also be triggered by drugs or other medical conditions. A psychotic episode does not by itself establish schizophrenia.
Prodrome
early phase of illnessA period in which signs of a developing disorder appear before a full psychotic episode. In the book’s account, a person may become withdrawn, anxious, disorganized, or troubled by mild suspicious thoughts while still retaining some awareness that something is wrong. Because cannabis use can increase during this period, researchers must consider whether it contributes to illness or follows its early symptoms.
Dose-response relationship
risk changes with exposureA pattern in which the likelihood or severity of an outcome varies with the amount, frequency, or duration of exposure to a substance. In this book, studies reporting greater psychosis risk among heavier cannabis users are treated as evidence supporting a causal role. A dose-response pattern strengthens an argument, but cannot by itself eliminate confounding or prove causation.
Cannabis use disorder
problematic cannabis useA clinical diagnosis describing cannabis use that becomes difficult to control and continues despite harm or disruption. The book uses the term when discussing people whose cannabis use is serious enough to be identified by clinicians, including in emergency-room and hospital records. It is distinct from occasional use and does not by itself establish that cannabis caused a person’s mental-health or behavioral problems.
Cannabis and Schizophrenia: A Longitudinal Study of Swedish Conscripts
Swedish cohort studyA 1987 study by Sven Andréasson and colleagues using records from Swedish men surveyed during military conscription in 1969 and 1970. Researchers compared reported cannabis use with later schizophrenia diagnoses. The study found higher risk among users, particularly frequent users, and adjusted for several possible confounding factors. It was an influential early study, but its observational design cannot alone prove causation.
Dunedin Multidisciplinary Health and Development Study
long-term New Zealand cohortA longitudinal study that followed a group of people born in Dunedin, New Zealand, in 1972 and 1973. Participants were assessed repeatedly from childhood into adulthood. Its measures included psychotic symptoms before many participants used cannabis, helping researchers examine whether early symptoms explained later associations between cannabis use and psychotic disorders. The study’s depth is a strength, while its relatively small sample limits some conclusions.
The Health Effects of Cannabis and Cannabinoids
National Academy evidence reviewA 2017 report by the National Academy of Medicine that reviewed thousands of studies on cannabis and cannabinoids. Its committee concluded that evidence supported a likely increase in schizophrenia and other psychosis risk, with greater use associated with greater risk. The report also assessed proposed medical benefits and harms, distinguishing stronger findings from claims supported by limited or inconsistent evidence.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.