FKBP51 compresses the functional range for T-allele carriers so severely that small cortisol fluctuations produce what feels like a binary switch between offline and fully on. This is the mechanism behind the co-occurrence of inattention and hyperfocus — they are below-range and above-range states of the same cortisol-glutamate system.
Inattention and hyperfocus co-occur because they are below-range and above-range states of the same cortisol-glutamate system. What follows is the proposed mechanism.
ADHD as currently diagnosed is a clustering of attentional and behavioral symptoms that can result from diverse underlying causes, rather than a single mechanistic entity. Given this heterogeneity, we do not make claims about ADHD broadly. Instead, we focus on two specific behavioral phenomena — inattention and hyperfocus — and propose a mechanistic model connecting FKBP5 to their co-occurrence. An association between FKBP5 polymorphisms and ADHD diagnosis has been established: Isaksson et al. (2015) found significant associations between FKBP5 variants, ADHD, and lower diurnal cortisol levels in children, and Kim and Jin (2021) confirmed this in Korean children. What has not been proposed, to our knowledge, is a mechanism explaining why these two seemingly opposite phenomena — inability to start anything and inability to stop — co-occur in the same individual.
FKBP51 compresses the functional range for T-allele carriers through two multiplicative effects. First, it shifts the cortisol dose-response curve rightward — more cortisol is needed to activate GR and drive glutamate release. Second, it strips AMPA receptors from synapses through the Hsp90 chaperone system (Part 1), meaning whatever glutamate IS released has fewer targets. The effective signal reaching downstream neurons is cortisol-driven glutamate release × AMPA receptor density, and FKBP51 reduces both. This double deficit makes the transition between "offline" and "online" occur over a much narrower band than in C/C carriers.
Neither of these baselines is fixed. The position of the cortisol dose-response curve and AMPA receptor density both reflect the cumulative effects of genotype, childhood calibration, recent stress history, seasonal light exposure, and current FKBP51 protein levels. Parts 3 through 5 describe how these baselines shift on different timescales. This section describes what happens minute to minute given wherever the baseline currently sits.
An important nuance: cortisol also binds mineralocorticoid receptors (MR) with roughly tenfold higher affinity than GR. MR is modulated by FKBP51 through the same FKBP51/FKBP52 competition mechanism, but the effect is weaker than on GR (see Part 1). This means the tonic MR-mediated signaling floor is partially degraded in T-allele carriers, not preserved. The compressed functional range is even narrower than GR dynamics alone would predict: FKBP51 reduces the ceiling (GR-dependent stress response) AND partially lowers the floor (MR-dependent tonic signaling). T-allele carriers with childhood adversity may have baseline cortisol low enough (see Part 5, "low-cortisol paradox") that even the partially attenuated MR is not fully saturated, which would compromise the tonic baseline further. The transition from "functioning but flat" to "fully online" occurs when cortisol rises enough to activate GR, glutamate release increases, and the glutamate signal exceeds what the AMPA-depleted synapses need to activate downstream dopamine, serotonin, and norepinephrine systems.
Most routine stimuli do not generate enough cortisol to reach this activation range in T-allele carriers, producing chronic understimulation. When a stimulus does push cortisol into range, glutamate floods synapses that have been running below capacity, dopamine neurons in the VTA come online, serotonin neurons in the raphe come online, and the individual shifts from offline to fully engaged. The sluggish FKBP5-mediated feedback means the system cannot disengage for hours. Ising et al. (2008) demonstrated this directly: healthy T/T carriers showed impaired cortisol recovery after a 15-minute psychosocial stressor, with cortisol staying significantly elevated compared to CC carriers throughout the post-stress observation window. The sustained cortisol dynamics of T-allele carriers may not be purely costly: more time in the GR-activated state may produce deeper processing, longer engagement with complex tasks, and solutions that a system with faster cortisol clearance would not reach.
But why does activation produce focused lock-in rather than diffuse arousal? The answer likely involves the locus coeruleus. Aston-Jones and Cohen (2005) showed that LC neurons operate in two modes: phasic (task-locked bursts producing focused attention) and tonic (elevated baseline firing producing distractible scanning). Grimm et al. (2024) confirmed with optogenetics that tonic LC stimulation engages exploratory processing while burst-like stimulation biases toward focused processing. The LC receives both direct GR modulation and glutamatergic excitatory input. Below the activation range, neither is adequate, and noradrenergic output defaults to tonic mode. When cortisol crosses into range, both GR modulation and glutamate-driven burst firing come online simultaneously — the individual locks onto whatever triggered the event, and sluggish FKBP51 feedback maintains this state for hours. An important constraint: local GABAergic interneurons prevent glutamate surges from becoming excitotoxic runaway cascades. The "compressed functional range" describes the transition between understimulation and adequate function, not between silence and seizure. GABA-mediated feedback inhibition scales locally to contain the glutamate surge within a functional band.
The mechanism described above is specific to T-allele carriers, but hyperfocus itself is not. At least two distinct mechanisms likely produce what is colloquially grouped under the same label.
Mechanism 1: Cortisol-glutamate gated activation (T-allele specific). This determines whether the individual can start. Cortisol must rise enough to drive glutamate past the AMPA-depleted threshold. Once it does, sluggish feedback sustains engagement. The canonical example: a person unable to initiate any productive activity all evening encounters an incorrect claim online and is now unable to disengage for three hours. The emotional provocation triggered a cortisol spike → glutamate surge → downstream systems came online, and FKBP5 feedback dynamics prevented shutdown.
Mechanism 2: Dopamine-timed reward cycling (genotype-independent). This determines whether the individual can continue once activated. Even above the cortisol-glutamate threshold, sustained engagement requires a fast enough feedback loop between action and reward. If the gap is short (writing code and seeing it work, a video game with instant feedback), dopamine sustains engagement indefinitely. If the gap is too long (a test suite that takes three minutes, a project where results are weeks away), the dopamine prediction signal decays before the reward arrives, and the brain disengages regardless of cortisol state.
Mechanism 1 is the ignition; Mechanism 2 is the engine. For T-allele carriers, Mechanism 2's reward-cycling machinery is inaccessible unless Mechanism 1 provides the cortisol-glutamate key. This dual model generates a specific prediction: T-allele carriers should show significantly more Mechanism 1 hyperfocus (triggered by provocation or intensity), while Mechanism 2 (dependent on feedback speed) should be distributed more uniformly across genotypes.
A T-allele child whose cortisol does not reach the activation range has underactivated GR in the locus coeruleus AND insufficient glutamatergic drive to LC neurons. The result is disinhibited, disorganized noradrenergic output: startle responses, difficulty sitting still, emotional overreactivity to minor provocations. This is the predictable consequence of a child whose activation range sits above what a quiet classroom provides. The same child in a high-engagement environment — sports, an exciting teacher, a novel experience — generates enough cortisol to cross into range, glutamate comes online, GR and glutamate both reach the LC, norepinephrine is properly modulated, and the child appears calm and focused. The inconsistency between contexts is the diagnostic clue: it is not the child who is inconsistent, but the cortisol-glutamate environment.
This reframes why stimulant medications reduce hyperreactivity in ADHD. Stimulants flood tonic dopamine directly but do not restore the glutamatergic context in which phasic dopamine signals are detected, nor do they address FKBP51-mediated AMPA receptor depletion. This may explain why stimulants produce function without full normalization — they compensate for the dopamine deficit without reversing the upstream glutamate deficit (Maltezos et al., 2014). Lithium orotate, by inhibiting GSK3β — simultaneously increasing tonic glutamate release and halting the kinesin-mediated removal of AMPA receptors — addresses a different layer of the problem entirely.
The FKBP5 model does not claim to explain all of ADHD. It explains one specific mechanism — cortisol-glutamate gated activation — which produces symptoms currently diagnosed under the ADHD umbrella. This mechanism interacts with other genetic variants to produce different presentations. Consider novelty-seeking, associated with the DRD4 7-repeat allele. In a CC carrier, high novelty drive is a mild personality trait. In a T-allele carrier, the compressed functional range converts it into what looks like pathological novelty dependence: without intense novel input, their glutamate never reaches the threshold dopamine neurons need to fire. DRD4 makes you want novelty; FKBP5 T/T makes you need it to function. FKBP5 determines how much input is needed to come online; other genes determine what kind.
Studies examining cortisol levels in ADHD populations have produced inconsistent results. This inconsistency is expected. The absolute cortisol number is not the relevant variable. Functional status is determined by at least four variables: (1) circulating cortisol level, (2) FKBP5 genotype, (3) childhood epigenetic calibration, and (4) current AMPA receptor density. A T/T carrier with a cortisol level of 15 mcg/dL and a CC carrier with the same level are in completely different functional states. Population studies that average cortisol across genotypes and childhood histories will inevitably produce noisy, contradictory results.