Personality Pharmacology

Part I · Foundations Draft

Chapter 6

The Personality-Control Theorem

A scientific theory usually earns credibility through data. Personality pharmacology is no exception. Claims about dopamine, endocannabinoids, psychiatric dimensions, or evolutionary selection all require empirical evidence. Yet the framework also contains a structural core that can be established with surprisingly little empirical detail.

It is useful to make that core explicit, because doing so separates what could be wrong at the edges from what would have to be wrong at the foundation.

Let x represent a set of brain variables relevant to personality. They might include circuit gain, reward sensitivity, threat weighting, inhibitory control, arousal, learning rates, or other parameters. Let c represent context and h the person’s developmental and learning history. Let P represent the person’s behavioral policy: the recurring way they attend, value, predict, feel, choose, and act.

Then we can write, schematically:

P = F(x, c, h)

Now introduce an intervention u. It may be a drug, sleep deprivation, exercise, meditation, stress, hormonal change, or another perturbation. The intervention changes the biological state according to some function:

x' = G(x, u)

The resulting policy becomes:

P' = F(G(x,u), c, h)

The notation is deliberately abstract. The important point is causal. If the intervention alters a brain variable and that variable influences personality-relevant behavior, then there is a pathway from intervention to behavior.

xbrain variablesPpersonality policyx′ = G(x, u)changed variablesP′changed policyFFintervention udrug, sleep, stress…context c and history hP′ = F(G(x, u), c, h)
Figure 6.1The causal path in one picture. An intervention reaches personality only through the brain variables it changes. Context and history shape the result at every step.

Five premises are sufficient.

First, personality-relevant behavior depends at least partly on brain function. This premise is difficult to deny while accepting that lesions, intoxication, anesthesia, sleep loss, neurodegeneration, hormones, or psychoactive drugs change behavior.

Second, people differ in at least some relevant brain variables. Human brains are not molecularly, developmentally, or functionally identical.

Third, at least some of those variables can vary quantitatively. Receptor density, release probability, learning rate, stress responsivity, transporter activity, or inhibitory gain need not come in only two discrete forms.

Fourth, psychoactive interventions alter some relevant variables.

Fifth, some altered variables influence behavior that belongs to the domain of personality, such as motivation, confidence, restraint, sociability, emotional intensity, vigilance, or cognitive flexibility.

If those premises hold, then personality pharmacology exists as a causal domain. The conclusion does not depend on whether dopamine predicts extraversion, whether traits are Gaussian, or whether the endocannabinoid model is correct. Those are questions about the map, not the existence of the landscape.

1Behaviour depends on thebrain2People’s brains differ3Some differences are amatter of degree4Drugs change some of thosevariables5Some of those variablesshape personalityPersonality-relevantdrug effectsmust exist, andmust differ fromperson to person
Figure 6.2The structural core. Five modest premises are enough to establish that personality-relevant drug effects exist and that they must differ from person to person.

A second result follows from heterogeneity.

Suppose a particular functional variable has an optimal region for a task. One person starts below it, another near it, and a third above it. A drug shifts the variable upward. The first person may improve. The second may change little or overshoot slightly. The third may worsen.

This is the simplest form of the baseline-dependence principle.

Interactive figure: Same drug, different starting points.

Figure 6.3Same drug, different starting points. Many functions work best in a middle range. Push three people by the same amount and the one who starts low improves, while the one at the sweet spot overshoots and the one already high gets worse. Illustrative model, not measured data.

In a multidimensional system the same logic becomes stronger. A drug may improve one dimension while worsening another. Increased motivation can be beneficial when inhibitory control is strong and hazardous when it is weak. Reduced threat may release a person from avoidance or remove caution they needed. Increased associative breadth may support creative generation or erode causal discrimination.

The optimal intervention is therefore conditional. We can write it schematically as:

u*(x,c) = argmin_u L(G(x,u), c)

The best intervention depends on where the system starts and what outcome matters in that context.

This yields an almost trivial but powerful theorem of personalization: if baseline information predicts which intervention produces lower expected loss, then a treatment policy that uses that information must perform at least as well as a policy that ignores it, and better whenever different subgroups truly have different optima.

The practical challenge is that useful baseline information may be difficult to measure. The logical result does not guarantee that a personality questionnaire, PET scan, genetic score, or blood test will supply it. Precision psychiatry can be theoretically correct and practically disappointing if the available measurements are noisy or the relevant dimensions remain hidden.

A third result concerns the notion of an ideal personality. Different tasks impose incompatible demands. Exploration competes with exploitation. Sensitivity competes with insulation. Rapid action competes with deliberation. Broad association competes with strict filtering. Trust competes with protection from exploitation. Present-focus competes with long planning.

ExplorationexploitationSensitivityinsulationRapid actiondeliberationBroad associationstrict filteringTrustprotection fromexploitationPresent focuslong planningEach task wants the knob somewhere else, so no single settingwins everywhere.
Figure 6.4No setting wins everywhere. Many demands on a mind pull in opposite directions, so the best position depends on the task.

If these trade-offs are genuine, there cannot be one fixed personality configuration that maximizes performance in every environment. The more general target is flexible access to different configurations.

A fourth result concerns psychopathology. If a psychological capacity is adaptive in moderation and harmful when severely excessive or deficient, then ordinary variation and at least some pathological states must occupy a shared dimension. Threat detection is the clearest example. Zero fear is not optimal; neither is constant fear. The same logic applies to reward pursuit, inhibition, attachment, vigilance, confidence, and associative generation.

This does not prove that every psychiatric disorder is dimensional. It proves only that the functions disorders disturb often are.

Finally, a normative inference follows, although it is not itself a scientific theorem. If a natural baseline is the contingent result of evolution and development, naturalness alone cannot establish that the baseline is morally optimal. Evolution selects for reproductive consequences under past environments, not for the reflective preferences of a present individual. A separate ethical argument is required before “natural” can become “ought”.

The reasoning in this chapter establishes the skeleton of personality pharmacology. It does not tell us which neurotransmitter controls which trait, how large the effects are, or which interventions are safe. Those questions remain empirical.

This distinction is protective. The theory should be willing to lose specific hypotheses without collapsing. The endocannabinoid archetypes may be wrong. A proposed serotonin-personality association may fail. A disorder believed dimensional may contain a genuine taxon. None of those findings would erase the more basic fact that heterogeneous biological systems generate heterogeneous responses to perturbation.

The architecture is robust. The map is negotiable.

With that separation in place, the next part begins the risky work of drawing the map.