Chapter 18: How to Read Interaction Patterns
Part III: Meaning and Interpretation
Research on couple conflict describes several forms of interaction. These descriptions can help a reader notice what occurs during disagreement, compare observations with the limits of a studied construct, and ask what remains unknown. They are not a catalog of kinds of people. They cannot tell a reader who a partner is, what that partner secretly intended, or whether a relationship is healthy or unsafe.
The distinction matters because a pattern name can be useful shorthand and a misleading label. Saying that a conversation contained demanding and withdrawing behavior directs attention to an interaction. Saying that one partner is “the demander” and the other is “the withdrawer” can make temporary positions sound like permanent identities. Use research terms to describe observable possibilities, not to assign fixed roles that people must occupy.
What an interaction description can tell you
Begin with a recognizable interaction form. Ask what the exchange looked like, how one response may have followed another, what each person reports experiencing, and what meanings they assigned to it. Then identify the limits of that description. An observable silence is not proof of indifference. A repeated question is not, by itself, proof of controlling intent. An explanation can be an attempt to clarify, defend, avoid responsibility, or simply supply information. The context and the person’s account matter.
Research on couple interaction can describe behaviors that occur together or in sequence, but a research measure may capture only part of an encounter. A laboratory discussion, daily report, and questionnaire do not necessarily describe the same phenomenon. In reading any research-described interaction, distinguish what was observed or reported from a possible interpretation. Consult the relevant research rather than treating a fictional example as evidence.
Consider a disagreement over a household expense. One person asks several questions about a purchase. The other begins defending the decision, then stops speaking. Someone observing the exchange might identify questioning, explanation, and disengagement. The observer cannot conclude from those behaviors alone that the first partner wanted control or that the second wanted to punish them. The original financial question may be legitimate. The interaction pattern and the underlying issue remain separate, and both may need attention.
Patterns are not personalities
Pattern names describe positions in exchanges. They do not establish psychological types, diagnoses, or enduring character traits. Even familiar terms from relationship research should be read this way. In a study of 182 married couples, observed demand-withdraw roles differed across discussion contexts (Eldridge et al., 2007). That does not mean every couple reverses roles or that context explains every role change. It does demonstrate why a single exchange should not be turned into a permanent portrait of either partner.
People may respond differently when a topic matters more to one partner, when the timing changes, or when their capacity to participate differs. One person may seek discussion about money and need distance during a discussion about relatives. The other may show the reverse. A research description identifies a possible interaction, not a rule that determines what either individual will do next.
Recognizing a pattern can improve the accuracy of a description. It should not become a new accusation. “I noticed that I kept pressing after you said you needed a break” is more informative than “You are always the avoidant one.” The first statement describes a potentially examinable event. The second turns a contested interpretation into an identity claim.
Patterns can overlap
Separately named research descriptions can help distinguish behaviors, but real exchanges do not have to fit neatly into those divisions. A conversation may include questions followed by defensiveness, criticism followed by countercriticism, and pressure followed by silence. These may be different views of one exchange rather than three independent patterns.
Overlap is especially important when one named interaction resembles another. Distinct terms do not automatically mean that two descriptions are scientifically distinct. Some labels refer to studied research constructs. Others are provisional educational groupings used to organize observations. The fact that an interaction is easy to recognize does not establish that it forms a separate, validated category.
A reader can consider several descriptions provisionally without choosing one “correct” diagnosis. What matters is whether a description helps identify actual behavior and clarify what remains uncertain. If no label fits, nothing is wrong with the observation. Research terminology is a limited aid to understanding, not an exhaustive map of every possible exchange.
The same issue is not always the same pattern
It is possible to revisit one issue through different interaction processes. A couple might discuss money calmly on one occasion, become stuck in repeated explanations on another, and withdraw from the conversation on a third. Conversely, a similar form of interaction may appear while discussing money, family obligations, or a change of plans.
These are distinctions to investigate, not guarantees of cross-topic consistency. The current research record does not establish that every couple repeats an identical sequence across topics or conflict episodes. Evidence about overall behavior, behavior during a single discussion, and recurrence across separate discussions must not be treated as interchangeable. Comparing encounters may be useful without claiming a universal recurring script.
When looking for repetition, record specific moments instead of relying only on a strong feeling of familiarity. What happened first? What did each person actually say or do? Did the conversation change direction? Was the same pattern present last time, or did the experiences merely feel similar? Answers may be incomplete. An accurate “I do not know” is more useful than supplying a motive or sequence that was never observed.
How to distinguish the evidence behind a description
The evidence basis matters as much as the label. A documented research pattern refers to an interaction form examined in external research, within the limits of its measurement, methods, and samples. Demand-withdraw is one example of a research tradition that has studied particular complementary behaviors. A named pattern is not necessarily universal, harmful in every context, or evidence that either participant has a fixed identity.
A research-informed synthesis combines findings or concepts to make an interaction easier to understand. The synthesis may be useful without having been evaluated as a distinct pattern. A BTL-specific construct is a term or organization belonging to Break the Loop, rather than a standardized scientific category. A provisional candidate is a proposed grouping whose conceptual distinctness or evidentiary basis still needs review.
These distinctions should not be collapsed. A research-described interaction pattern such as demand-withdraw identifies complementary behaviors or patterns of responding; it is not itself a BTL Loop. Within BTL, the proposed recurring pattern is the progression Launch → Overload → Old Story → Protection Move, even when the topic, emotion, particular Old Story, or Protection Move differs between occasions. Evidence of criticism, withdrawal, emotional arousal, or attribution does not independently validate that full sequence. This is an educational organization, not a research finding that all couples follow four identical stages. When applying BTL, identify only the stages responsibly supported by the person’s account; leave unknown stages unknown.
The distinction also applies to statistical classifications. In a pooled observational analysis of 1,957 mixed-gender couples, researchers identified multiple communication profiles from coded laboratory discussions. One profile with relatively high negative reciprocity also had relatively high reported relationship satisfaction (Wojda-Burlij et al., 2026). This is a limited descriptive finding, not evidence that negativity is desirable. More importantly, profiles derived from one ten-minute discussion are not permanent couple identities or demonstrations of repeated sequences across daily life.
What a description cannot establish
A disciplined account can ask about the observable form, possible ordering of responses, each person’s reported experience, relevant meanings, research constructs, and what not to assume. Those questions are prompts for inquiry, not a validated classification method. Even when one response appears to follow another, the description does not establish that a specific sequence is causally reinforced across later conflicts.
Consider common responses, contexts in which an exchange occurred, related research constructs, and places where a different choice might be available. Within BTL, some responses may be examined as Protection Moves, but an apparent function is a question to investigate, not a justification for harm. A person who withdraws may be overwhelmed, disengaged, attempting to punish, setting a legitimate boundary, or taking a needed pause. The same visible movement can have different meanings and consequences.
The question “What should I not assume?” is central. A description cannot establish motive, personality, causation, or equal responsibility where that evidence is absent. References should identify research relevant to the specific claim. Fictional examples illustrate how the ideas might appear; they are not proof that a described sequence is common or inevitable.
A brief way to examine an interaction
Begin with a disagreement you can describe without inventing anyone’s motives. Write down the issue and then the interaction: what was said, what changed, and what followed. If a research-described form seems relevant, compare the observation with the construct’s actual definition, methods, and limitations. A resemblance alone does not establish a diagnosis or the recurrence of a full BTL Loop.
Next, consider differences as carefully as similarities. Did one person set a clear boundary rather than simply withdraw? Did the same person occupy a different position on another occasion? Is the original concern still unresolved? If you have only one person’s account, acknowledge that the other person’s experience remains unknown. The goal is to make observation more precise and assumptions less automatic.
If BTL language is useful, map only what can responsibly be identified: a possible Launch, signs of Overload, a familiar Old Story, and a Protection Move. Do not force all four positions into every exchange. The broader educational goal is to recognize a response with enough clarity to consider another choice, not to produce a perfect diagram of the relationship.
Safety and responsibility come first
Describing interaction patterns is not the primary approach for understanding coercive control, intimidation, threats, violence, stalking, sexual coercion, fear of retaliation, or related abuse. In those circumstances, shared-pattern language can obscure power differences and wrongly suggest equal responsibility. Safety, accountability, and appropriately specialized support take priority. The person experiencing harm should not be asked to make an unsafe conversation work by finding a better response in the Loop.
Even outside those situations, a mutual sequence does not imply that each partner contributed equally. One person’s behavior can be more harmful, or one partner can have a legitimate boundary or grievance. Recognizing interaction does not erase responsibility for the behavior itself. Nor does choosing to end an unsafe or unproductive conversation automatically amount to an unhealthy pattern.
The purpose of recognition
Research terms and BTL vocabulary can support examination of an exchange, not a verdict about a person or relationship. Pattern names should remain revisable descriptions. Explanations should remain distinguishable from observations. Examples should remain examples, and claims about evidence should say exactly what the research supports.
A useful examination ends with more precise questions: What did I observe? What am I inferring? What matters about the underlying issue? Remains unknown? What response, if any, is available to me without minimizing harm or surrendering a boundary? Those questions are more valuable than finding a label that appears to explain everything.
Chapter 19 turns from research-described interaction forms to money as a substantive issue. A specific topic can matter on its own terms even when a BTL Loop may also unfold around it.
Back to the Field Guide table of contents
Previous Chapter 17: How Two Loops Become One Recurring Pattern