Relationship memories sometimes produce logs that look completely broken:
“They were a good person.” “Some things they did still annoyed me.” “I still picture their face.” “I miss having them around.” “I have absolutely no desire to get back together.” “I cared about them.” “But the future package around the relationship did not work for me.”
A binary interface cannot handle this.
Do you love them? Yes / No.
Do you want them back? Yes / No.
Were they a good person? Yes / No.
If humans were that simple, nobody would be awake at 2 a.m. trying to understand a breakup.
The problem is not necessarily contradictory emotion. The problem is often bad data modeling.
1. If “love” and “go back” share one variable, everything looks corrupted
Imagine a former partner.
You may still value them highly as a person. Attachment can remain. Nostalgia can remain. Some anger can remain. The shared history can still matter. You may still have zero intention of restarting the relationship. And when marriage, cohabitation, family systems, money, work, or lifestyle enter the picture, long-term compatibility may still be poor.
Compress all of that into a single “love score” and the result becomes nonsense.
Store it in separate columns and it becomes ordinary:
- attachment: high
- anger: partly present
- evaluation of the person: positive
- sense of loss: present
- intention to reunite: none
- long-term compatibility: low
- willingness to accept surrounding life conditions: none
“I miss them, but I would not go back” is not a logical contradiction.
Missing someone and choosing a relationship are different fields.
2. Apparently opposite emotions can coexist
Psychology studies this under terms such as mixed emotions and ambivalence.
A meta-analysis of 63 experimental studies found that mixed emotional states can be reliably elicited and are not merely a measurement artifact [1].
Romantic relationships show the same pattern. Research published in 2024 examined partner ambivalence as a state in which positive and negative thoughts and behaviors toward the same partner can coexist [2].
So:
“I loved them” and “I am not going back”
do not have to defeat each other.
They may simply answer different questions.
Humans are more database than toggle switch.
3. A four-gate model can reduce decision accidents
A useful practical framework is to separate processing layers instead of trying to erase feelings.
Emotion gate
What do I feel?
Love. Sadness. Anger. Relief. Nostalgia. Tenderness.
Do not ask yet whether the feeling is rational.
Facts and logic gate
What actually happened? Are the recurring problems real? Do the long-term conditions fit? What are the realities of money, family, work, housing, and daily life?
Feeling lonely does not automatically improve compatibility.
Interpretation gate
What does the feeling mean?
Thinking about someone does not mean I should return.
Being angry does not mean the entire person was bad.
Ending a relationship does not mean the shared years had no value.
Do not manufacture causality just because the emotion is strong.
Authority gate
Who gets to decide what?
The other person decides their life. You decide your non-negotiable conditions. Shared life decisions require shared agreement. You do not control another family system. You also do not have to enter it.
This gate helps separate “I do not condemn you” from “I consent to this future.”
The four gates are not literal neuroscience. They are a bookkeeping metaphor for cleaner decisions.
4. “Women are complicated” is a weak explanation. Humans are complicated
People often joke that women have complicated feelings.
Science does not support the cartoon version in which women possess complex emotion and men are simple machines.
Mixed emotions are studied as a general human phenomenon [1]. Findings about sex and gender differ depending on which emotional skill or regulation strategy is measured, as well as culture and context.
A more useful everyday distinction may be this:
not who has more feelings, but who separates and verbalizes more of them.
Once the full record is spoken aloud, it sounds dramatic:
“I loved them.” “I would not return.” “They were good to me.” “Some conditions were unacceptable.” “I miss them.” “My decision is unchanged.”
It looks like a complicated romantic monologue.
It may simply be uncompressed data.
5. Putting feelings into words does not delete them. It helps separate them
Recent work describes affect labeling as a process of accumulating evidence from ambiguous internal information until a person can decide what they are feeling [3].
Emotion differentiation refers to how precisely people distinguish and describe their emotional experiences. A 2024 experience-sampling study tested whether scaffolding emotion-language use could influence this differentiation [4].
The useful target is not emotional deletion.
You do not need to erase: “I miss them.” “I am angry.” “I loved them.”
You can instead stop the automatic conversions:
I miss them → therefore I should go back.
I loved them → therefore I must accept every future condition.
I am angry → therefore the entire person was terrible.
Store the emotion. Process the decision separately.
6. If “men must endure” compresses everything into one ZIP file, the data can leak elsewhere
Many men encounter norms such as:
“Do not be weak.” “Take it like a man.” “Stop talking about feelings.”
It would be sloppy to turn that into “men suppress emotions, therefore men drink.”
A 2024 meta-analysis of 35 studies and 32,412 adult men found only a modest average association between conformity to masculine norms and alcohol outcomes, around r = .17 [5].
Separately, a meta-analysis of 43 papers found greater emotional suppression was associated with poorer social well-being, including lower social support, social satisfaction, and romantic relationship quality [6].
A more careful story is:
multiple emotions exist → expression may be restricted → some people compress them into “I’m fine,” “whatever,” work, isolation, drinking, or other outlets.
The internal data may not be simple.
The logging system may just be disabled.
7. You do not need one clean story in order to make a clean decision
Humans like narratives with one cause:
“They were bad, so I left.” “I still love them, so I should return.” “If I will not return, I never loved them.” “If I am angry, the whole relationship was a mistake.”
These stories are tidy.
Tidy is not the same as accurate.
You can preserve the messy version:
“They were a good person.” “Some things hurt.” “I still miss them.” “The relationship mattered.” “I am not going back.” “Our long-term conditions do not fit.”
Then make one behavioral decision.
That is not indecision.
It is separation between the emotional database and the decision engine.
Conclusion: humans are not one-bit systems
Love/hate. Return/leave. Good/bad.
Those switches are too small for human relationships.
Attachment, anger, gratitude, grief, tenderness, and incompatibility can all exist toward the same person. A person can preserve all of those feelings and still decide not to restart the relationship.
Research on mixed emotions supports the coexistence of apparently opposing states [1][2]. Work on affect labeling and emotion differentiation offers ways to describe that internal complexity more precisely [3][4].
For practical purposes:
feelings are feelings, facts are facts, interpretations are interpretations, authority is authority, and the decision comes last.
“I loved them. I still would not go back.”
Not a bug.
A feature.
References (6)
- Berrios R, Totterdell P, Kellett S. “Eliciting mixed emotions: a meta-analysis comparing models, types, and measures.” Frontiers in Psychology. 2015. PMID: 25926805 pubmed.ncbi.nlm.nih.gov
- Zoppolat G, Righetti F, Faure R, Schneider IK. “It’s complicated: The good and bad of ambivalence in romantic relationships.” Emotion. 2024. PMID: 38300552 pubmed.ncbi.nlm.nih.gov
- The process of affect labeling.” 2026. PMID: 41109773 pubmed.ncbi.nlm.nih.gov
- Effects of scaffolding emotion language use on emotion differentiation and psychological health: an experience-sampling study.” Cognition and Emotion. 2024. PMID: 39048111 pubmed.ncbi.nlm.nih.gov
- Zamboanga BL, Newins AR, Ford K, Perrotte JK. “Conformity to masculine norms and its association with drinking behaviors and alcohol-related consequences among adult men: A meta-analysis.” Social Science & Medicine. 2024. PMID: 39303377 pubmed.ncbi.nlm.nih.gov
- Chervonsky E, Hunt C. “Suppression and expression of emotion in social and interpersonal outcomes: A meta-analysis.” Emotion. 2017. PMID: 28080085 pubmed.ncbi.nlm.nih.gov
