Can CBD Prevent Persistent THC Memory Deficits Despite Higher Plasma THC? A Mouse Study

Short answer: In one controlled male-mouse experiment, giving cannabidiol (CBD) alongside delta-9-tetrahydrocannabinol (THC) prevented the particular THC-associated behavioral differences detected after three weeks of injections. Yet in a separate group of adolescent mice, directly assayed plasma THC was approximately 30% higher with concurrent CBD than with THC alone at a single sampling time, 24 hours after the final injection (p = 0.027). This is a useful warning against claiming that a behavioral difference must reflect a lower circulating parent-drug concentration. It is not evidence that CBD reliably prevents memory impairment in humans, that higher plasma THC is harmless, or that an individual driver’s THC measurement predicts behavior.

These results come from Murphy and colleagues’ complete 2017 original primary experiment, Chronic Adolescent Δ9-Tetrahydrocannabinol Treatment of Male Mice Leads to Long-Term Cognitive and Behavioral Dysfunction, Which Are Prevented by Concurrent Cannabidiol Treatment, Cannabis and Cannabinoid Research. The original study is available at DOI 10.1089/can.2017.0034. The scientific question is deliberately narrow: can a coadministered drug alter selected experimental outcomes even when measured parent THC in the sampled plasma compartment does not decrease?

Why can the plasma-and-behavior disagreement matter in forensic toxicology?

Forensic toxicology distinguishes a measured chemical concentration from a person’s functional condition. An analytical result describes the analyte, instrument, specimen and collection time. A behavioral experiment describes observed performance on a defined task. Neither measurement automatically supplies the other. The Murphy experiment is informative because the intervention that accompanied improved group task performance also accompanied higher, not lower, plasma THC in a separate pharmacokinetic sampling experiment.

That disagreement does not mean the plasma result was wrong or that THC had no behavioral effects. Rather, it shows why the simplistic inference “less apparent impairment means less circulating THC” was not supported under these precise experimental circumstances. A plausible pharmacodynamic interaction was discussed by the investigators, but they did not establish its molecular mechanism by experimentally isolating a receptor pathway. Nor did they measure the brain’s THC concentration or a full post-dose plasma concentration-time curve.

In a human case, this distinction helps frame the proper questions. Was the reported number from venous whole blood, separated plasma, serum or another matrix? What specific parent compound or metabolite was quantified? Was functioning actually measured? Was CBD exposure established and at what time? No mouse group average can replace those case-specific observations.

What original study design produced the comparison?

The investigators used male CD1 mice. Their Methods section reports 221 mice overall, while the published main behavioral schedule assigns 52 adolescent-treatment mice and 51 adult-treatment mice to five named groups. Those numbers must not be confused. The schedule’s group denominators are directly presented for the displayed main behavioral cohorts; separate analyses, including plasma measurements, do not provide a single common denominator that can be inferred from the 221 total.

One set of animals received daily intraperitoneal injections across adolescent postnatal days 28–48; another was treated during early adulthood, postnatal days 69–89. Each treatment phase lasted 21 days. The five assigned groups were untreated cage controls, vehicle injections, THC 3 mg/kg/day, CBD 3 mg/kg/day, and combined THC 3 mg/kg/day plus CBD 3 mg/kg/day. The combined regimen therefore delivered equal nominal masses of the two cannabinoids—not a commercial cannabis product, a fixed inhaled dose, or a clinical prescription.

These 52 and 51 mice are not equivalent to 103 separately collected plasma samples. Figure 1 lists them in the behavioral schedule; repeated tests after the same treatment do not create independent animals. A distinct adolescent plasma-analysis cohort was reported without a figure-stated independent group count. Treating technical mass-spectrometry injections, repeated behavior sessions or a later result as a new independent subject would incorrectly inflate evidentiary precision.

When were behavior and drug concentrations actually assessed?

Behavioral testing began one day after the last cannabinoid injection and was repeated following a 42-day drug-free interval. The first battery proceeded across several days: open field, novel-object recognition, marble burying, elevated-plus maze and Nestlet shredding. The second battery used the same general outcomes weeks later. These are organized rodent behavior tests, not a human driving examination or a standardized clinical assessment of a patient.

The plasma experiment was different. Investigators collected plasma from a separate group of adolescent male mice after 21 days of THC alone or THC plus CBD. Collection occurred 24 hours after the last injection. Thus the displayed plasma finding is neither a blood draw from every behavior-tested animal nor a simultaneous pharmacokinetic measurement at each behavioral endpoint. It cannot be paired to individual discrimination-index scores or used to estimate each mouse’s concentration while it performed the delayed test.

What exactly did the novel-object recognition task measure?

Novel-object recognition (NOR) exploits a mouse’s tendency to examine a novel object longer than a familiar one. Animals first explored two equivalent objects for five minutes. After a 105-minute interval, they were returned for a three-minute trial in which one object had been replaced. The observer scored exploration when the animal’s nose pointed toward an object within the specified short distance. Familiar and novel object placement was counterbalanced.

The study calculated a discrimination index: (time exploring novel object − time exploring familiar object) divided by (the sum of those two exploration times). The index can range from −1 to +1. A positive number reflects greater relative examination of the novel object; zero indicates no preference under that metric. It is a defined rodent object-exploration measure, not an estimate of a human person’s recall accuracy, legal competence, psychosis or overall intelligence.

Three weeks of THC alone decreased the discrimination index in both adolescent- and adult-treated groups immediately after treatment. After the six-week interval, the lower index persisted in the adolescent-treatment group but was not maintained in the adult-treatment group. The authors reported no major treatment-group differences in total exploration time that would account for the principal discrimination-index pattern. Nevertheless, novelty preference can involve motivation and perceptual attention, so a single score is not a pure molecular readout of memory.

Did concurrent CBD prevent the measured recognition deficit?

Within the study’s specific comparison, yes: the groups coadministered equal-dose CBD and THC did not show the THC-only reduction in the discrimination index. That attenuation was reported for both age-at-exposure schedules and for immediate testing; the adolescent delayed-testing result was especially important because the THC-only difference remained after 42 days, whereas the concurrent group did not display the same persisting deficit. CBD alone did not materially change the principal NOR result.

The proper wording is “the selected THC-associated group difference was prevented in this experiment,” not “CBD has been clinically proven to cure THC memory loss.” The investigators did not randomize human patients, observe a driver’s lane position, study multiple CBD:THC ratios, test every route of use or establish a general protective dose. They tested a 1:1 administered mass ratio in mice, and only for the reported conditions and outcomes.

What were the directly measured plasma THC results?

Unlike the behavioral rating scores, the plasma measurement is an actual chemical analysis. The laboratory prepared a lipid extract from sampled mouse plasma, using a deuterated anandamide internal standard as described by the authors. The extract was analyzed with high-performance liquid chromatography coupled to an Applied Biosystems API 3000 triple-quadrupole tandem mass spectrometer, using multiple-reaction monitoring of parent THC’s 315.2 precursor-to-123.2 fragment transition.

Figure 8 reports that adding CBD raised measured parent plasma THC by approximately 30% compared with THC-only treatment at the 24-hour point. The figure labels the between-group difference p < 0.027; the Results paragraph reports p = 0.027. The graph’s vertical scale is nanomolar; the primary source does not report precise group means with confidence intervals or the independent number of mice per plasma group in that figure. It would be improper to invent exact measured values, a human ng/mL whole-blood cutoff or a validated behavioral prediction coefficient from its bar heights.

Crucially, this is separated mouse plasma parent THC, not a direct forensic venous whole-blood specimen. The study did not assay the same mice’s blood during the delayed NOR task. It did not report oral-fluid THC, urine THC-COOH, hair testing, vitreous concentration, or a breath sample. Nor did it measure a full concentration-time course, THC brain concentrations or serial active 11-hydroxy-THC exposures sufficient to explain all pharmacokinetic possibilities.

What positive and nonsignificant comparisons must be kept together?

Study interpretation changes when nonsignificant findings remain visible alongside positive ones. The authors used one-way analysis of variance, Tukey follow-up comparisons where appropriate and two-group Student tests, with a nominal significance threshold of 0.05. The published behavioral graphs mostly supply significance markers rather than exact per-contrast p values and do not supply a confidence interval for each comparison. Those limitations should be reported instead of converting marker symbols into fabricated decimal values.

Why does the marble-burying result deserve particular care?

The authors described an immediate adolescent trend toward burying more marbles after THC. But their explicit comparison against vehicle was p = 0.06, above the stated 0.05 significance threshold. Some plotted group comparisons to CBD-containing conditions could differ, yet those are not the same hypothesis as THC versus vehicle. The study did not find a lasting marble-burying difference among adolescent injection groups after six weeks.

A forensic report that quotes this task as unqualified proof of significantly increased adolescent THC-induced compulsivity relative to vehicle would be inaccurate. Likewise, the adult CBD-alone group buried fewer marbles on the immediate test, but the reported differences disappeared at the delayed test. Multiple endpoints and comparison groups require test-specific descriptions, not a blanket assertion that every repetitive-behavior assay changed in the same direction.

What does Nestlet shredding actually tell us?

Nestlet shredding quantifies how much cotton nesting material a mouse tears up during a standardized 75-minute observation. The authors interpreted the result as a repetitive or compulsive-like behavior in rodents. Adolescent THC-alone dosing increased shredding both soon after treatment and after the six-week interval. Adult THC exposure did not yield the analogous significant increase against vehicle. Equal-dose concurrent CBD attenuated the adolescent THC pattern in this experiment.

Shredding is observable physical behavior, not a direct cannabinoid assay and not a clinical diagnostic interview. Altered nesting could have several motivational or behavioral dimensions. It would overreach to label a particular human patient as having obsessive-compulsive disorder because laboratory mice shredded more Nestlet material after a defined injection sequence.

What about delayed anxiety-like behavior?

On the elevated-plus maze, delayed testing after THC-only exposure showed fewer entries into and less time in open arms for both adolescent- and adult-treated animals. The concurrent CBD-treated groups did not show the same THC-only pattern. Immediate differences were not identical across all ages, and the paper separately reported no consistent changes in total maze-arm entries. Accordingly, an argument that the finding was simply grossly suppressed locomotion would need evidence beyond these measurements.

Other anxiety-related measures did not agree uniformly. The open-field assay found no significant between-group differences in total distance traveled or in thigmotaxis. Its several-minute acclimation period also limited what it could establish about initial novelty responses. The correct description is a selective elevated-plus-maze outcome, not proof that the tested drug regimen caused generalized anxiety in every behavioral paradigm or in any human individual.

Does higher plasma THC prove that CBD was pharmacodynamically protective?

No. The finding is inconsistent with a simple claim that the observed attenuation resulted from a lower measured parent THC concentration in plasma at that one time point. A pharmacodynamic interaction is plausible. The authors discussed CBD’s potential negative allosteric modulation of cannabinoid CB1 receptors as a candidate explanation, not as a mechanism directly established by an antagonist experiment within this study.

Many alternative details remain unmeasured or uncertain. A single 24-hour sample cannot supply peak THC concentration, integrated exposure, tissue distribution, the exposure during an earlier task or metabolite profiles. Plasma and behavior were measured in different animal groups; consequently, there is no within-animal relation showing that the mice with the highest plasma THC had the best NOR performance. A higher late plasma value could coexist with a different early concentration curve, altered distribution, altered active-metabolite formation or other pharmacokinetic changes. Those possibilities cannot be excluded by this experiment alone.

The more defensible conclusion is methodological: an observed drug concentration and an observed functional response are two distinct endpoints, and their relationship requires direct validation. Background on those issues appears in the site’s biological-matrices guide and pharmacokinetics guide.

Can an injected 1:1 THC/CBD ratio be generalized to cannabis products?

It cannot be treated as a universal product standard. The mice received purified cannabinoid quantities injected intraperitoneally, with THC and CBD each at 3 mg/kg daily. Smoking, vaping, oral oil, edible products and topical exposures have different absorption and first-pass processes. Products sold as balanced or CBD-rich can differ in absolute dose, cannabinoid composition, contaminants, dosing frequency, use history and analytical labeling. Even a verified 1:1 ratio on a retail label would not establish the same systemic exposure or developmental timing as these scheduled mouse injections.

Nor should the result be conflated with other original studies of adolescent female rats receiving different CBD-rich/low-THC mixtures. The present experiment tested male mice, an equal-dose regimen and a separate direct parent-THC plasma assay. Different species or sex, compound ratios, developmental windows and assays can lead to different results without making either primary paper interchangeable. Any synthesis across them should explicitly retain their different independent DOI records, conditions and endpoints.

What does the experiment not establish about schizophrenia?

The authors discussed schizophrenia-related features because object-recognition differences, anxiety-like testing and repetitive behavior appear in relevant basic-neuroscience models. Nevertheless, the study did not diagnose schizophrenia in an animal or observe psychotic illness in a person. No human patient cohort was administered THC or CBD under this protocol, and no clinical diagnostic endpoint was measured. Mouse NOR and Nestlet tests cannot function as patient-level diagnostic criteria.

Claims about the epidemiology of cannabis and psychiatric disorders in the paper’s introduction involve cited other studies, not outcomes independently generated by the Murphy experiment. They should not be presented as if this mouse experiment itself estimated a human psychosis risk ratio or proved prevention by a particular product. A result can inform a mechanistic hypothesis while remaining unable to establish a particular illness or cause in litigation.

How would a case-specific toxicology review apply this paper?

First identify the actual proposition offered. If the proposition is that equal-dose CBD accompanied attenuation of selected THC-related rodent behaviors even though late mouse plasma parent THC was higher, the paper directly bears on it. If the proposition is that a named driver with a given whole-blood THC concentration was or was not impaired, the paper provides no validated individual classifier.

Second identify the physical specimen and analytical method. The study’s chemical observation was parent THC in separated mouse plasma using LC–MS/MS, not an immunoassay surrogate or legally obtained human whole blood. The 24-hour sampling time and separate cohort should accompany any quotation of the approximately 30% increase. For a real forensic specimen, confirm analyte identity, collection timing, calibration, matrix-specific validation and quality controls. The site’s confirmatory-testing guide addresses that evidentiary distinction.

Third preserve actual denominators and negative findings. The main behavioral arms were between eight and twelve animals for the adult groups and between nine and eleven for adolescent groups; the paper’s 221 total does not turn every endpoint into 221 independent measurements. Report the p = 0.06 marble-burying null, nonsignificant open-field endpoints, and recovery of adult-treated NOR after delay alongside the positive adolescent persistence. Finally separate group-level animal observations, pharmacological hypotheses and specific event causation. None can be silently substituted for the others.

Bottom line: behavior and THC concentration are not interchangeable

The central result is narrower—and more useful—than a universal assurance about CBD. In male CD1 mice given 21 days of injected THC, selected object-recognition and behavior differences followed THC-alone treatment, particularly when exposure began in adolescence. Concurrent equal-dose CBD attenuated the deficits measured under this protocol. But a separate adolescent plasma cohort had more parent THC after concurrent CBD at the one measured time, not less. This paradox cautions against single-number impairment reasoning while leaving the molecular mechanism and every human case-specific conclusion unproven.

Frequently asked scientific and forensic questions

1. Did CBD reduce THC’s measured plasma concentration in this study?

No. The original adolescent-mouse plasma experiment reported an approximately 30% increase in parent THC with concurrent CBD at 24 hours after the last 21-day injection sequence, with p = 0.027. It did not report a full concentration-time curve.

2. Was the plasma comparison performed in the same animals as the memory tasks?

No. The paper explicitly describes a separate adolescent mouse group for plasma sampling. A group-level plasma difference cannot be paired with each memory-tested mouse’s performance.

3. Did THC-related novel-object differences persist after treatment stopped?

They persisted after the 42-day drug-free interval in adolescent-treatment mice receiving THC alone. The adult-treated group did not show the same persistent NOR decrease, despite an immediate effect.

4. Does the result establish that CBD protects adolescent human memory?

No. It establishes selected mouse-group findings after scheduled intraperitoneal doses. There was no controlled human exposure study, human memory diagnosis or validated protective dose.

5. Was marble burying significantly greater after adolescent THC versus vehicle?

Not for the expressly reported immediate THC-versus-vehicle contrast: p = 0.06, which did not meet the paper’s 0.05 criterion. Other between-group comparisons cannot be substituted for that one.

6. Is nanomolar mouse plasma THC equivalent to a human whole-blood forensic result?

No. The matrix, species, assay, sampling schedule and pharmacokinetics differ. The primary experiment did not determine a conversion formula or any impairment threshold for human whole blood.

7. Does this study prove the mechanism or cause of a person’s psychiatric or driving event?

No. A pharmacodynamic CBD–THC interaction was hypothesized but not isolated mechanistically here. No person, on-road driving behavior, clinical diagnosis or legally relevant incident was investigated.

Original primary source: Murphy M, Mills S, Winstone J, Leishman E, Wager-Miller J, Bradshaw H, Mackie K. Chronic Adolescent Δ9-Tetrahydrocannabinol Treatment of Male Mice Leads to Long-Term Cognitive and Behavioral Dysfunction, Which Are Prevented by Concurrent Cannabidiol Treatment. Cannabis and Cannabinoid Research. 2017;2(1):235–246. https://doi.org/10.1089/can.2017.0034.