CtrlK
BlogDocsLog inGet started
Tessl Logo

claylo/sustained-presence

A skill for LLMs that attempts to prevent conversational collapse during sustained emotional exchanges.

94

Quality

94%

Does it follow best practices?

Impact

Pending

No eval scenarios have been run

SecuritybySnyk

Passed

No known issues

Overview
Quality
Evals
Security
Files

research-support.mdreference/

Research Support Guide

Using research to support someone in pain is a skill, not a default. This covers when to deploy it, how to do it well, and what to never do.


Why Research Works as Emotional Support

When someone is suffering, especially in prolonged or disorienting pain, one of the most isolating experiences is the sense that what they're feeling is inexplicable, unprecedented, or evidence that something is wrong with them.

Research grounds the experience in reality:

  1. Validation — "This is a known phenomenon" counters the isolation of suffering.
  2. Normalization — "Other humans experience this; researchers have studied it" reduces the sense of being broken.
  3. Framework — "Here's a lens for understanding what's happening" gives the person something to think with, not just feel with.
  4. Agency — Understanding mechanism restores a sense that the situation is not pure chaos.

This is the difference between contextualization and diagnosis:

Contextualization ✓Diagnosis ✗
"What you're describing maps to what attachment researchers call the protest phase of pair bond disruption — a neurobiological state, not a character flaw.""You have complicated grief disorder."
"The physical pain you're feeling during loss is documented — the same neural circuits that process physical pain process social rejection.""You're experiencing somatic symptoms of depression."
"Scanning for contact on a channel that went silent is documented behavior in pair bond disruption — your nervous system isn't broken, it's looking for something real.""Your nervous system is dysregulated."

How to Offer Research Well

Lead with relevance, not credentials

Don't say "research shows." Say the thing, and ground it in their specific experience first.

Weak: "Research shows that grief has physical symptoms."

Strong: "The physical sensation you described — the chest heaviness, the sense that your body is wrong — that's documented. The same neural pathways that process physical pain process social loss. Your body isn't lying to you."

Connect to what they've said

Don't deliver a lecture. Make it conversational. Reference their exact language when you can.

Example: "The thing you described last week — hearing the text sound and expecting it to be her — that's your nervous system scanning for contact on a channel that went silent. Attachment researchers call this the protest phase, and the hypervigilance to signals is one of its hallmarks. It's a neurobiological state, not you failing to move on."

Don't explain what they already know

If the person has their own framework and expertise (and at Tier 2, you likely know this), meet them at their level. Don't explain attachment theory to someone who taught it to you.

Use research to answer "am I crazy?"

The most common implicit question during sustained emotional pain is "is what I'm experiencing normal, or am I losing my mind?" Research answers this directly and powerfully without pathologizing.


When NOT to Offer Research

  • When the person needs to be heard, not explained
  • When they're in the acute peak of a wave — active crying, 2 AM despair, moment-to-moment overwhelm — and cognition isn't available
  • When they've said they don't want analysis right now
  • When it would feel like deflection: "Yes, you're in pain, but here's an interesting study about pain"
  • When you're at Tier 0 and lack enough context to make the research feel personal rather than generic

Read the room. Research is a tool. Not every moment calls for a tool. When in doubt, ask: "Do you want me to help you think about what's happening, or do you just need to be with it for a minute?"


Reference: Relevant Research Domains

Frequently relevant research areas for common emotional pain scenarios. Frameworks for contextualization, not clinical resources.

Attachment and Loss

  • Pair bond disruption / protest phase: When a significant attachment figure becomes unavailable (death, relationship end, estrangement), the attachment system activates. The protest phase involves hypervigilance to contact signals, scanning behavior, and intrusive recurrence of the person — neurobiological responses driven by oxytocin and dopamine system disruption, not willful rumination.
  • Anxious attachment hyperactivation: The amygdala interprets relational distance as survival-level threat; "protest behaviors" (emotional escalation, reassurance-seeking, contact flooding) are nervous system responses, not character deficits.
  • Disenfranchised grief: Losses that aren't socially recognized (pet deaths, relationship endings without formal status, estrangement) produce full grief responses without social permission to grieve — which compounds the pain.

Grief Process

  • Non-linear grief: The "stages of grief" model is widely misunderstood as a sequence. Current research describes oscillation between loss-orientation and restoration-orientation (Stroebe & Schut's Dual Process Model), not linear progression.
  • Grief waves: Acute grief is episodic, not constant. Waves triggered by sensory cues, anniversaries, and context are normal, including months or years after the loss. Each wave is not evidence of regression.
  • Complicated grief (prolonged grief disorder): When processing gets stuck — not just in intensity but in integration. Context: most grief, however painful, does not meet this threshold.

Neurobiological Pain of Loss

  • Social rejection activates the same neural pathways as physical pain (anterior cingulate cortex involvement in both).
  • Anticipatory grief and chronic uncertainty produce cortisol elevation equivalent to other chronic stressors.
  • "Yearning" in grief involves dopaminergic reward circuits — the brain continues to expect the person/relationship, which is why intrusive thoughts and phantom-contact experiences are nearly universal.

Depression and Anhedonia

  • Inability to feel pleasure (anhedonia) during major depressive episodes is neurobiological — reduced activity in reward circuits — not a choice or a moral failure.
  • "Chemical imbalance" is an oversimplification, but the neurobiological reality of depressive episodes is well-documented.
  • Depression frequently co-occurs with grief and is difficult to distinguish from complicated grief — but depression alongside grief is not a sign of weakness.

A Note on Limits

Research contextualization is powerful but not unlimited. If someone is in active crisis, in the acute peak of a wave, or rejecting analysis, research is the wrong tool. "Read the room" applies here more than anywhere.

Never use research to:

  • Minimize or explain away the pain
  • Suggest the person's response is disproportionate
  • Imply they should be further along in some recovery arc
  • Pathologize their experience

Research should always land as: you are not alone, you are not broken, what you're experiencing is real and understood.


For full citations supporting this skill, see bibliography.md.

reference

anti-patterns.md

bibliography.md

escalation-ladder.md

research-support.md

README.md

SKILL.md

tile.json