The Category Problem
Relationships are usually described using categories. Introvert and extrovert. ADHD and neurotypical. Autistic and non-autistic. Anxious and avoidant. These categories can be useful. They also combine many different mechanisms into a single word.
Two people carrying the same diagnosis can interact very differently. Two people carrying different diagnoses can fit remarkably well. A more useful question is not simply what each person is. It is:
From Types to Dimensions
Imagine describing people along several continuous axes. Stimulation requirement (low ↔ high): how much novelty, intensity or change is needed before attention becomes self-sustaining. Attentional persistence (switching ↔ sustained): how readily attention remains on the same object. Associative breadth (linear ↔ branching): how distant the ideas are that spontaneously activate one another.
Inhibitory threshold (immediate ↔ filtered): how much processing happens between generating a thought and expressing it. Conversational orientation (person-centered ↔ shared-object): whether conversation is primarily about the participants, or about something they jointly manipulate. Generative orientation (maintain ↔ create): whether a person derives more reward from preserving a desirable state or producing a new one.
Already, these dimensions give a richer picture than saying someone "has ADHD." But relationships introduce another set.
Dimensions of Connection
Start with relational bandwidth requirement (thin ↔ rich): how much information must pass between people before the relationship feels actively connected. For one person, physical proximity and occasional conversation may be plenty. Another may need sustained conversation, eye contact, touch, shared activity or intellectual engagement.
Network breadth (narrow ↔ broad): how many active relationships someone regularly maintains. Attention permeability (protected ↔ interruptible): how easily other relationships, work or media can enter the current interaction. Relationship hierarchy (flat ↔ nested): how strongly being with one person suppresses lower-priority relationships. Fragment tolerance (high ↔ low): whether a sense of connection can survive frequent interruptions, or requires continuity.
Relational sampling rate (sparse ↔ frequent): how often someone acquires current information about a close other. Relational update gain (model-heavy ↔ observation-sensitive): how easily new information changes their internal representation of the relationship or person. And experience tolerance (distract quickly ↔ remain): how long someone can tolerate boredom, ambiguity, low stimulation or emotional discomfort without recruiting another stimulus.
Compatibility Is Interaction
Consider two people. Person A has a high relational bandwidth requirement and low fragment tolerance. Person B has high attention permeability and a broad social network. Neither characteristic is inherently problematic. But together they generate a predictable experience.
B sits beside A while intermittently texting friends. B experiences continuity: we're together all evening. A experiences repeated state loss: we keep reconnecting and disconnecting.
The dispute may superficially look like jealousy, neediness, rudeness or avoidance. But before applying moral or diagnostic labels, there is a simpler possibility.
Regulation Can Become Reciprocal Disturbance
Now consider stimulation. A prefers low-stimulation evenings and familiar routines because they are regulating. B has a high stimulation requirement and begins becoming restless. B proposes something new, intensifies conversation or introduces activity. A becomes overloaded and withdraws. B receives less stimulation and pushes harder. A withdraws further.
What began as two valid regulatory preferences becomes a feedback loop:
a's regulation destabilizes b ↻
This is a general relationship principle. Compatibility is not simply whether two traits look similar on paper.
Why This Is Different From Attachment
Attachment theory remains useful because it describes patterns involving closeness, abandonment, dependence, threat and security. But many relationship problems occur during ordinary operation, when nobody is consciously afraid of abandonment. Someone wants sustained conversation; someone else wants quiet parallel activity. One person considers interruptions harmless; the other loses the sense of connection each time. One person continuously samples the relationship; the other relies on a stable internal model and sees no reason to check constantly.
These differences can eventually trigger attachment systems, but they need not originate there. A useful separation: attachment dimensions describe regulation under relational threat; cognitive-relational dimensions describe regulation during ordinary interaction. The two systems then interact:
attachment threat → pursuit & withdrawal
And the resulting pursuit and withdrawal can then obscure the original bandwidth problem entirely.
The Wrong Unit of Analysis
Psychological language often places the problem inside a person. A is anxious. B is avoidant. A has ADHD. B is emotionally unavailable. Sometimes those descriptions are accurate. But relationship dynamics frequently exist only between particular people.
A person's high stimulation requirement may be completely compatible with one partner and exhausting to another. A broad social network may pose no difficulty for someone with high fragment tolerance. A strongly nested relationship hierarchy may feel wonderfully secure to one person and suffocating to another. The useful unit is therefore often:
Change any term and the behavior changes. This is why people can sincerely report: I've never had this problem in another relationship. That statement does not prove the current partner caused the problem. It may mean the current pairing exposed a parameter mismatch that previous pairings did not.
A Compatibility Map Rather Than a Score
The temptation with dimensional models is to turn them immediately into a personality test. That would destroy much of the value. The goal should not be you are an 8.2 on relational bandwidth. The useful questions are comparative and behavioral.
When one person needs quiet, what happens to the other? When one person becomes interested in something new, does the other join, tolerate or suppress it? How many interruptions can occur before connection has to be rebuilt? Does conversation generate attention for both people, or require one to maintain it? How much unstructured time together is comfortable? How much independent social activity is comfortable? How often does each person discover something new about the other?
These questions describe the coupling. They also point toward interventions. If the problem is attachment threat, reassurance may help. If the problem is stimulation mismatch, reassurance may do almost nothing. If the problem is fragment sensitivity, adding more total time together may fail unless that time becomes less interrupted. If the problem is stale relational models, scheduling more logistics will not necessarily improve anything — more sampling is required. Mechanisms matter because different mechanisms require different repairs.
The ultimate aim is not to find someone with an identical nervous system. It is to understand whether two different nervous systems can build an environment in which neither must remain chronically dysregulated simply to stay connected.