Start here

What this page is, and what it isn’t

Some cravings and low moods are not about willpower. They track a physical pattern — what time of day it hits, what you reach for, what happens when you skip a meal. Learning to read that pattern is useful whether or not you ever take a supplement.

This page describes a way of grouping those patterns into five types. The framework is the work of Julia Ross, MA, MFT, a licensed therapist who spent decades directing outpatient eating disorder and addiction programs in the San Francisco Bay Area (Ross, 2018). She set it out across three books — The Diet Cure (Ross, 1999), The Mood Cure (Ross, 2002), and The Craving Cure (Ross, 2017). Everything below is our summary of her model, in our words, with links to her originals.

Her framework is a clinical observation system, not a validated medical test, and parts of the science behind it have not held up. We say which parts below, because you deserve to know that before you act on any of it.

This page does not tell you to take anything. Orchard Human Services provides counseling and clinical training. We do not prescribe, recommend, or dose supplements, and nothing here is medical or nutritional advice.

Safety first

Talk to a prescriber before you take anything

If you are in crisis, or if you are having thoughts of harming yourself, please contact a crisis line or emergency services rather than a supplement. In the US, call or text 988.

The five patterns

Reading your own pattern

The five types below are Ross’s (2017), described here in our own words. Most people recognize themselves in more than one, which she treats as the norm rather than the exception. The scale under each one shows how many items on her questionnaire belong to that type, and where her threshold falls (Ross, n.d.).

TYPE 1

The afternoon slide

Serotonin

Cravings arrive as the light goes — late afternoon, evening, or in the middle of the night. Winter is worse than summer. Alongside the cravings there is often worry, rumination, self-criticism, perfectionism, irritability, or trouble getting to sleep and staying asleep.

Often reached for: sweets, bread and other starches, alcohol, chocolate, cigarettes (Ross, 2021).

17 questionnaire itemsscore above 7

Ross also asks you to rate each checked item 0–10; she treats most ratings above 3 as meaningful (Ross, n.d.).

TYPE 2

The empty-tank crash

Blood glucose

The trigger is the gap between meals. Skip breakfast and by mid-afternoon you are shaky, headachy, unable to concentrate, and snapping at people. Cravings spike specifically because you went too long. There may be a family history of hypoglycemia, diabetes, or alcoholism (Ross, 2017).

Often reached for: sugar, starches, caffeine, alcohol (Ross, 2021).

11 questionnaire itemsscore above 4

This is the pattern most often mistaken for an anxiety or anger problem. If it fits you, meal timing is worth discussing with a doctor before anything else.

TYPE 3

The comfort reach

Endorphins

Certain foods are not just liked — they are loved, and they function as company. Chocolate, bread, cheese, ice cream, and combinations of dough and dairy sit at the top. There is often real sensitivity to emotional and physical pain, easy tears, a history of chronic pain or unresolved trauma, and a strong pull toward anything that numbs or soothes.

Often reached for: chocolate, bread and pasta, dairy, alcohol, painkillers, cannabis — and behaviors that work the same way, including hard exercise (Ross, 2018, 2021).

14 questionnaire itemsscore above 6

Ross (2018) reports this is the highest-scoring of the five among people seeking help for food cravings. It overlaps heavily with grief, pain, and trauma histories, so it is worth bringing to a therapist, not only to a doctor.

TYPE 4

The wound-up body

GABA

The anxiety lives in the body rather than the head: stiff or aching muscles, an inability to loosen up, a cluttered mind, a sense of being close to overwhelm. Stillness is hard — meditation, prayer, or simply resting feels effortful rather than restorative. Snacking is a way to discharge tension.

Often reached for: alcohol, cannabis, cigarettes, sweets and starches, anti-anxiety medication (Ross, 2021).

11 questionnaire itemsscore above 4

Ross (n.d.) lists difficulty meditating, praying, or being still among the markers here. If you have been told to “just meditate” and found it impossible rather than merely difficult, this pattern is worth taking seriously.

TYPE 5

The flat battery

Dopamine & norepinephrine

Not sad so much as switched off. Low drive, low motivation, boredom, difficulty holding attention. The depression here is flat rather than agitated. Sweets and caffeine work as a temporary charge rather than as comfort.

Often reached for: coffee, energy drinks, soda, chocolate, sugar, stimulant medication or drugs (Ross, 2021).

8 questionnaire itemsscore above 4

Ross (2002) flags that this pattern overlaps closely with an underactive thyroid, which is a medical question with a blood test attached to it.

The instrument

Take the questionnaire at the source

Ross publishes her questionnaire free (Ross, n.d.). We link to her original rather than hosting a copy, so you are always working from the current version as she wrote it, and so she receives the traffic and the credit for her own instrument.

Being straight with you

Where the science is thinner than the books suggest

We think you should have this before you spend money on supplements.

• The core theory is contested. The idea that depression and anxiety are caused by low neurotransmitter levels — the premise the whole framework rests on — was not supported by a large systematic umbrella review (Moncrieff et al., 2023). The symptom groupings may still be useful; the explanation for them probably is not right.
• GABA supplements may not work the way they are described. The evidence that swallowed GABA reaches the brain is contradictory at best, and the mechanism behind any effect remains unresolved (Boonstra et al., 2015; Hepsomali et al., 2020). People do report effects; the reason is unclear.
• Tryptophan and 5-HTP have thin trial support. A review of 108 studies found only two of sufficient quality, covering 64 people in total. The direction was favorable, but the evidence was judged too weak to be conclusive (Shaw et al., 2002).
• Tyrosine works in a narrow window. It helps under acute stress, sleep loss, or heavy cognitive load, and mostly does not at baseline (Jongkees et al., 2015).

None of this means the framework is useless. It means it should be treated as a set of ideas to test carefully with your own doctor, one change at a time — not as a protocol to follow.

What newer research adds — including in Ross’s favor

Research published since her books has clarified two things she got broadly right and one thing her model leaves out entirely.

• Food may matter more than capsules. A controlled trial that changed only dietary tryptophan — the food, not a supplement — found better mood, less depression, and less anxiety on the higher-tryptophan diet (Lindseth et al., 2015). And tryptophan eaten in whole foods, alongside fiber and plant compounds, appears to act on the body through routes a purified capsule does not (Chehadi et al., 2026). Ross’s insistence on eating more real protein looks better supported than her supplement list.
• The serotonin picture is more nuanced than “no evidence.” Lowering dietary tryptophan reliably worsens mood in people who have been depressed before or have depression in the family, while doing little in people who never have (Jenkins et al., 2016). Serotonin does appear to matter — but as a vulnerability that interacts with other things, not as a simple deficiency.
• Sleep may be the real mechanism — and it is the best-supported part of all this. Tryptophan’s clearest effect is on sleep: falling asleep faster, sleeping longer, waking less. And improving sleep genuinely improves mental health — a review of 65 clinical trials involving over 8,600 people found meaningful reductions in depression, anxiety, rumination, and stress, with bigger sleep gains producing bigger mood gains (Scott et al., 2021). Sleep is also when the brain reprocesses emotional experience (Walker & van der Helm, 2009) and when most growth hormone is released, supporting repair (Van Cauter et al., 1998). Whether tryptophan reaches your mood through sleep has not actually been tested — but it is a more solid route than the deficiency story.
• Inflammation can redirect the whole thing. Most tryptophan never becomes serotonin; it goes down a separate route called the kynurenine pathway. Inflammation and chronic stress actively push it that way (Chehadi et al., 2026; Jenkins et al., 2016). This is missing from Ross’s model, and it may explain why some people follow her framework carefully and feel nothing.

Practical version: if this framework fits you but supplements have done nothing, the answer may not be a higher dose. It may be a question about inflammation, stress, gut health, or sleep — and that is a conversation for a physician.

And if you take one thing from this page, take this: working on your sleep is better supported by evidence than anything in the supplement list above, costs nothing, and is something a therapist can help you with directly. That is a reasonable place to start.

Next step

Bring the pattern, not the protocol

If something here fits you, the most useful thing you can do is write down the specifics — what time of day, what you reach for, what happens when you skip a meal, what the craving does for you — and bring that to your next appointment. That description is clinically useful to almost anyone you show it to.

Orchard Human Services provides counseling and clinical supervision. We can work with you on the behavioral and emotional side of these patterns and coordinate with your medical providers. We do not prescribe or dose supplements.

Credit where it belongs

About this framework and its author

Julia Ross, MA, MFT is a licensed psychotherapist who combined roughly 40 years of clinical practice with three decades of pioneering work applying nutritional therapy to mood problems, eating disorders, and addictions. She founded several integrative treatment programs in the San Francisco Bay Area beginning in 1980 and later directed a virtual clinic for food cravers, and she trains health professionals through the Neuro-Nutrient Therapy Institute (Ross, 2018).

The five-type model, the questionnaire, the cut-off scores, the symptom categories, and the amino acid pairings on this page are all hers. Orchard Human Services has restated them in our own words for accessibility and added an independent appraisal of the research evidence. That appraisal is ours alone and should not be attributed to Ross.

If this framework is useful to you, please support her work directly by buying her books and using her own tools rather than relying on secondhand summaries, including this one.

References

Provenance

What this model is, and where it came from

Julia Ross, MA, MFT, directed outpatient eating disorder and addiction programs in the San Francisco Bay Area beginning in 1980, combining roughly 40 years as a licensed psychotherapist with three decades of nutritional therapy work (Ross, 2018). Her framework derives from high-volume clinical observation — by her account, over 20,000 individual amino acid trials since 1996 (Ross, 2018) — not from controlled research.

The model matured across three books. The Diet Cure embeds amino acids inside an eight-step functional program (Ross, 1999). The Mood Cure extracts the neurotransmitter piece and makes it the whole intervention, using four “false mood types” (Ross, 2002). The Craving Cure reframes the enterprise as addiction medicine and adds a fifth target: blood glucose availability (Ross, 2017).

Ross’s distinction between true emotions — proportionate responses that resolve — and false moods arising from biochemical depletion is the conceptual foundation of the whole system (Ross, 2002). Adding glucose as co-equal with the four neurotransmitters is arguably her strongest structural insight (Ross, 2017, 2018). Neurotransmitter synthesis and release are ATP-dependent; a client in a glycemic trough will present with symptoms mimicking all four neurotransmitter profiles at once. Ross treats hypoglycemia as its own type rather than folding it in.

Her clinical adoption of amino acid therapy was prompted by Kenneth Blum’s Reward Deficiency Syndrome work in the 1980s and 1990s (Blum et al., 1988; Ross, 2002, 2018). She situates the food-addiction parallel within contemporary reward neuroscience, citing Volkow and Baler (2015). That lineage matters for evidence appraisal and is addressed below.

Master map

System, presentation, and intervention

All content in this table is Ross’s, consolidated from Ross (2017, 2018, 2021). Scroll horizontally on narrow screens.

Three details that secondary summaries usually lose

• Endorphins need broad dietary protein, not a single precursor. Ross (2018) specifies at least four ounces of complete animal-source protein per meal for low-endorphin clients, because endorphin peptides require up to 19 amino acids to build, whereas serotonin and dopamine each need one. A free-form single amino cannot construct a 19-residue peptide. This is the most clinically actionable nuance in the framework and is routinely dropped from secondary summaries.
• DPA is a degradation blocker, not a building block. D-phenylalanine is proposed to inhibit enkephalinase, slowing endorphin breakdown rather than supplying substrate (Ehrenpreis, 1985; Ross, 2018). It is mechanistically unlike every other amino in the system.
• Ross rejects the fixed tryptophan-to-tyrosine ratio rule. Her position is that the two should generally be dosed at different times of day and titrated independently against symptoms, not co-administered in a set ratio (Ross, 2018).
• GABA is her low-dose outlier. Ross (2018) reports stronger effects from 125 mg GABA than from much larger doses of other aminos, with many clients remaining at the starting dose; she reports disliking 750 mg products.

Substance cross-map

Reading substance choice as functional self-medication

This view is the one most useful in substance use settings, and it survives independently of whether any supplement is ever recommended. Asking what the substance is doing for the client produces better conceptualization than reading use as pathology alone. The mappings below are consolidated from Ross (2017, 2018, 2021).

The endorphin row is the widest, and Ross (2018) reports that endorphin scores are typically the highest of all five categories among food-craving clients even when several categories are elevated. If you are setting a screening priority, that is where the base rate sits. Note that Ross also treats gliadin (gluteomorphin) and casein (casomorphin) as opioid-active food peptides, with fat and salt as milder opioid stimuli (Ross, 2018).

Scoring

Cut-offs and assessment method

Scoring is two-stage: check each item that describes a typical day (one point each), then rate checked items 0–10 for severity. Totals are plotted across all five types to produce a profile (Ross, n.d.). The instrument evolved from a four-part checklist in The Mood Cure (Ross, 2002) to the five-part 0–10 severity-rated version in The Craving Cure (Ross, 2017).

Sequencing rule

Ross’s clinic addresses the highest-scoring symptom first, with one stated exception: when catecholamine and serotonin deficits co-occur, correct catecholamines first, on the reasoning that raising serotonin in an unstimulated system produces spaciness and further attentional loss (Ross, 2002).

Amino trialing

Her distinctive procedure, added to standard assessment in 1996, is single-capsule in-office trialing at the lowest available dose, observed over minutes, with a second dose trialed on non-response; she attributes major compliance gains to the immediacy of the feedback (Ross, 2018). Her stated rescue for an adverse response is 1,000–2,000 mg oral vitamin C powder in four ounces of water (Ross, 2018) — an unvalidated clinic practice with no published mechanism, which should not be represented to anyone as an antidote.

Laboratory testing

Ross (2018) explicitly rejects urinary neurotransmitter testing as clinically unreliable, noting that results frequently contradict clear symptom pictures and that supplements prescribed from urine results often produce no effect or harm. She reports preferring blood platelet testing, and secondarily plasma, where corroboration is wanted. Worth emphasizing in training, because these panels are heavily marketed to integrative practitioners — and here Ross is correct, against her own commercial interest. Peripheral urinary monoamine metabolites do not index central neurotransmitter tone.

Expected duration

Weeks to months for children; months to a year for adolescents and adults; longer where there is a family history of alcohol or drug addiction, which Ross treats as a marker of genetic neurotransmitter dysregulation (Ross, 2018). Discontinuation is predicated on a sustained high-protein, nutrient-dense diet being in place.

Safety

Contraindications

All contraindications in this section are Ross’s, consolidated from her published precautions chart (Ross, 2021), except where noted as Orchard’s commentary.

Two issues Ross underweights

Appraisal

Grading the claims

This is what converts the framework from a belief system into a usable clinical tool. Teach the grades alongside the model, not after it. The appraisal in this section is Orchard’s and does not represent Ross’s positions — the claims in the left column are hers; the grades and the reasoning in the right column are ours.

What we teach interns to keep

• The five-type typology as a phenomenological map. Distinguishing cognitive worry from somatic tension, and anergic from negativistic depression, is useful differentiation whether or not the neurochemistry is as described.
• Blood glucose as co-equal with the neurotransmitters. Symptoms arising from a glycemic trough are routinely misattributed to anxiety, irritability, or emotional dysregulation.
• The functional reframe of substance choice.
• The protein co-requirement for peptide neurotransmitters.
• Skepticism toward urinary neurotransmitter panels.
• The protein co-requirement, which newer work justifies more broadly than Ross does (Chehadi et al., 2026).
• Her attention to sleep — which the independent literature supports more strongly than any other element of her framework, and which sits inside counselor scope (Scott et al., 2021).

And what to discard

• The deficiency-causes-disorder mechanism as a general model.
• The GABA blood-brain barrier claim as stated.
• The Blum-derived efficacy statistics.
• The questionnaire’s status as an assessment.
• The permissiveness around serotonergic drug interactions.
• The assumption that precursor intake predicts neurotransmitter output, with no immune or metabolic gate in the model.

The missing variable

Inflammation, and the kynurenine shunt

Ross’s model has no inflammation variable anywhere in it. Two decades of tryptophan research say that omission matters more than any dosing detail on this page.

Only a small fraction of dietary tryptophan ever becomes serotonin. The kynurenine pathway accounts for roughly 90% of tryptophan metabolism (Jenkins et al., 2016). Which branch the substrate takes is not fixed — it is actively regulated. Pro-inflammatory cytokines including IL-6, TNF-α, and IFN-γ induce indoleamine 2,3-dioxygenase, and cortisol stimulates tryptophan 2,3-dioxygenase; both divert tryptophan away from serotonin synthesis and toward kynurenine and its downstream neurotoxic metabolite quinolinic acid (Chehadi et al., 2026).

The clinical consequence is direct. A client carrying a high inflammatory or chronic-stress burden converts a smaller share of what they take, and may need more substrate — or less inflammation — to reach the same effect. Under Ross’s framework that reads as a failed amino trial, and the next move is a dose increase. Under a kynurenine-aware framework it reads as a substrate being shunted — and the intervention target is the inflammation, not the precursor. Elevated kynurenine-to-serotonin ratios have been linked to melancholic features, treatment resistance, and anhedonia (Chehadi et al., 2026).

Kałużna-Czaplińska et al. (2019) reach a similar conclusion from the nutrition side: tryptophan availability is bound up with brain–gut axis function and immune regulation, not with intake alone. Jenkins et al. (2016) add that only about 5% of the body’s serotonin is made centrally, with enterochromaffin cells in the gut epithelium accounting for roughly 90% of total synthesis, and that gut microbiota directly regulate how much tryptophan is available in the first place.

A second correction: food is not the same as a capsule

Ross’s protein co-requirement for endorphin clients turns out to have broader justification than she gives it. Chehadi et al. (2026) draw an explicit distinction between purified tryptophan or 5-HTP supplements, which mainly raise systemic substrate availability, and tryptophan delivered in whole-food matrices alongside fiber, resistant starch, and polyphenols — which additionally act on the upstream regulatory nodes, downregulating IDO1 and supporting tryptophan hydroxylase activity. On that reading, a free-form amino capsule and a tryptophan-dense meal are not interchangeable interventions, and the meal has mechanisms the capsule lacks.

The strongest single piece of support for the dietary route is Lindseth et al. (2015): a randomized crossover trial in 25 healthy young adults comparing four days of a high-tryptophan diet (10 mg/kg/day) against four days at the RDA (5 mg/kg/day), with a two-week washout. The high-tryptophan diet produced significantly more positive affect (p < .01), fewer depressive symptoms, and less anxiety (p < .05). Negative affect did not differ significantly. That is a small, short, healthy-volunteer study — but it is a controlled dietary trial, and it points at food rather than at a bottle.

Where this leaves Ross

Ross’s afternoon-and-evening dosing convention for serotonin, her tryptophan preference in children, and her protein emphasis all survive this literature reasonably well — tryptophan has consistent effects on sleep latency, total sleep time, and night waking, and is described as usable in pediatric populations where standard hypnotics are inappropriate (Yousef et al., 2024). What does not survive is the implicit assumption that precursor in equals neurotransmitter out. The conversion is gated, and the gate is immune and metabolic.

The strongest chain

Sleep as the downstream mechanism

There is a route by which tryptophan could plausibly support mood and cognition that does not depend on the serotonin-deficiency argument at all — and it is better evidenced than the argument Ross actually makes.

The reasoning runs: tryptophan’s best-supported single effect is on sleep. Sleep is causally upstream of mental health. Therefore tryptophan may reach mood through sleep rather than by directly correcting a neurotransmitter deficit. Each link in that chain has independent support.

Link 1 — sleep is causally upstream of mental health, not merely correlated

This is the load-bearing citation. Scott et al. (2021) meta-analyzed 65 randomized controlled trials comprising 72 interventions and 8,608 participants, asking whether interventions that improve sleep also improve mental health. Improving sleep produced a medium effect on composite mental health (g+ = −0.53), with depression (−0.63), anxiety (−0.51), and rumination (−0.49) all showing medium effects, stress small-to-medium (−0.42), and positive psychosis symptoms small (−0.26). Critically, they found a dose–response relationship: greater sleep improvement produced greater mental health improvement. That pattern is what distinguishes a causal relationship from a correlational one.

For scale: a g of −0.63 on depression from improving sleep is a larger and far better-evidenced effect than anything claimed anywhere else on this page.

Link 2 — overnight clearance (contested, and worth teaching as contested)

Xie et al. (2013) reported that natural sleep or anesthesia in mice produced roughly a 60% increase in interstitial space, driving convective exchange of cerebrospinal and interstitial fluid and increasing β-amyloid clearance — the finding that made the glymphatic system famous.

Link 3 — REM and emotional memory

Walker and van der Helm (2009) set out a REM-sleep hypothesis of emotional-memory processing: overnight modulation of affective neural systems reprocesses recent emotional experience and recalibrates next-day reactivity of limbic and autonomic networks. They propose this as a brain-based account of why sleep abnormality both initiates and maintains mood disturbance. It is a heavily cited model rather than a settled finding, but it is the most coherent available explanation for why poor sleep and mood disorder track each other so tightly — and it is directly relevant to trauma work, where the emotional-reprocessing function is the point.

Link 4 — growth hormone, with a precision correction

Growth hormone is secreted in pulses, and in normal young men a major secretory episode occurs shortly after sleep onset in temporal association with the first period of slow-wave sleep; roughly 70% of daily GH output occurs during early sleep, with a linear relationship between the amount of slow-wave sleep and concurrent GH secretion. In women the sleep-dependent contribution is lower and more variable (Van Cauter et al., 1998).

The correction worth carrying into practice: GH release is tied to slow-wave sleep, not REM. The two mechanisms therefore live in different halves of the night — slow-wave sleep dominates the first half, REM the second. Anything that fragments early sleep costs GH; anything that truncates the night costs REM. This is why “get more sleep” and “go to bed earlier” are not interchangeable pieces of advice, and it is a useful distinction to teach.

Where the chain breaks

Every link above is separately evidenced. The chain as a whole has never been tested end to end. No study has followed tryptophan supplementation through to improved sleep architecture through to clearance, GH output, or REM-dependent emotional processing, and out to mood outcomes. Presenting the chain as demonstrated would repeat exactly the error this page criticizes Ross for — a plausible mechanism narrated as an established pathway. It is a strong rationale for investigation and a reasonable clinical hypothesis. It is not a finding.

Gaps

What has not been done

• No independent replication of Ross’s clinic outcomes by unaffiliated investigators.
• No psychometric validation of the Craving Type or Mood Type questionnaires.
• No randomized trial of the five-amino protocol as a package, in any population.
• No trials of DPA or DLPA in modern opioid use disorder populations, despite a plausible mechanism and urgent clinical need.
• No head-to-head comparison against N-acetylcysteine, which has the stronger craving evidence base (Winterlind et al., 2024) and works through cystine-glutamate exchange — a route absent from Ross’s five-system model entirely.
• No published safety surveillance from the 20,000+ trials Ross (2018) reports conducting.
• No study stratifying amino acid response by inflammatory burden or kynurenine-to-tryptophan ratio — the most obvious candidate explanation for non-response in her framework, and testable.
• No head-to-head comparison of free-form amino supplementation against a matched high-tryptophan whole-food diet, despite mechanistic reasons to expect they differ (Chehadi et al., 2026; Lindseth et al., 2015).
• No study tracing tryptophan through sleep architecture to mood outcomes — the mediation analysis that would test whether the sleep route explains any of the effects Ross attributes to direct serotonergic action.

Sources

Attribution and primary materials

References