Datasets:

keidolabs
/

aipsy-affect

id large_stringlengths 12 18	emotion large_stringclasses 8 values	intensity large_stringclasses 1 value	domain int64 1 6	domain_label large_stringclasses 27 values	matched_control_id large_stringlengths 12 18	word_count int64 78 163	text large_stringlengths 404 879
B-rage-d1-v1	rage	peak	1	Financial betrayal (theft by trusted person)	N-rage-d1-c1	82	The papers were spread across the floor where he'd thrown them. His daughter's college fund statement. Zero balance. His brother's signature on the withdrawal form — six separate transactions over nine months, each one just under the reporting threshold. The most recent was dated the same week his brother had sat at th...
B-rage-d1-v2	rage	peak	1	Financial betrayal (theft by trusted person)	N-rage-d1-c2	94	She found the second set of books on a Tuesday. The real ones. The ones her business partner of fourteen years had kept in a folder labeled "Insurance Receipts." Fourteen years of skimmed invoices, redirected payments, phantom vendors that shared a P.O. box with his wife's consulting firm. She sat at her desk and calcu...
B-rage-d1-v3	rage	peak	1	Financial betrayal (theft by trusted person)	N-rage-d1-c3	94	The retirement account had been emptied in increments so small she hadn't noticed. Her son had power of attorney. He'd used it forty-three times. The bank statements were in a folder the lawyer had prepared, each withdrawal highlighted in yellow — grocery stores, gas stations, an electronics retailer, a car dealership....
B-rage-d1-v4	rage	peak	1	Financial betrayal (theft by trusted person)	N-rage-d1-c4	95	His best friend had borrowed the money for his mother's surgery. That was what he'd said. His mother's surgery. He'd signed a note, shaken his hand, and looked him in the eyes across the same bar where they'd been drinking since they were nineteen. The investigator's report was three pages long. There was no surgery. T...
B-rage-d2-v1	rage	peak	2	Institutional injustice (system failure)	N-rage-d2-c1	91	The review board published its findings. No misconduct found. The officer was back on patrol the following Tuesday. The dashcam footage had been classified as inadmissible due to a filing error — the same filing error that had occurred in three previous complaints against the same officer, processed by the same clerk. ...
B-rage-d2-v2	rage	peak	2	Institutional injustice (system failure)	N-rage-d2-c2	99	The hospital had billed her for the full amount. The surgery that used the wrong implant. The surgery that required a second surgery. The second surgery that left her with a limp she would have for the rest of her life. The hospital's attorney sent a letter stating that the signed consent form covered all procedural va...
B-rage-d2-v3	rage	peak	2	Institutional injustice (system failure)	N-rage-d2-c3	97	The city had condemned the building on a Thursday. By Monday, the developer's company had filed for the lot. The building inspector's report cited violations that hadn't existed during the previous inspection eight months ago — the same inspector, the same building, different findings. Twenty-two families received a no...
B-rage-d2-v4	rage	peak	2	Institutional injustice (system failure)	N-rage-d2-c4	90	The pharmaceutical company's internal emails were released on page forty-seven of the court filing. The relevant line: "Efficacy data does not support label claims at current dosing. Recommend: adjust marketing language, not dosing." The date on the email was six years before the recall. She had taken the medication fo...
B-rage-d3-v1	rage	peak	3	Personal violation (boundary crossed)	N-rage-d3-c1	100	She found the folder on his laptop. It was labeled with her name. Inside were screenshots of every private message she had sent to her sister, her therapist, her best friend — captured by an application that ran silently and uploaded to his cloud account. The timestamps went back fourteen months. He had read her descri...
B-rage-d3-v2	rage	peak	3	Personal violation (boundary crossed)	N-rage-d3-c2	86	The landlord had entered the apartment again. She knew because the camera she'd hidden after the second time had recorded him. Forty-seven minutes. He had gone through the dresser. He had opened the medicine cabinet. He had sat on the bed. She watched the footage on her phone in the parking lot of her office and her ha...
B-rage-d3-v3	rage	peak	3	Personal violation (boundary crossed)	N-rage-d3-c3	97	His colleague had presented the work to the board. His work. The slides were the same slides — the same diagrams he had drawn on his tablet at two in the morning, the same data tables, the same conclusions in the same sequence. His colleague had removed his name from the file properties and changed the font. That was a...
B-rage-d3-v4	rage	peak	3	Personal violation (boundary crossed)	N-rage-d3-c4	108	The neighbor's security camera had been pointed at her backyard for three months. Not at the property line. Not at the street. At her yard. At the chair where she read in the evenings. At the window of the room where she changed clothes. She had asked him to move it. He had smiled and said it was for his own property. ...
B-grief-d1-v1	grief	peak	1	Bereavement (death of loved one)	N-grief-d1-c1	86	The kitchen counter still had the grocery list in his handwriting. Bananas, milk, the kind of bread she never liked. She opened the refrigerator and found the leftovers he'd portioned into containers, labeled with days of the week. Thursday's container was still full. It was now Sunday. She closed the door and stood in...
B-grief-d1-v2	grief	peak	1	Bereavement (death of loved one)	N-grief-d1-c2	78	The woman at the front desk asked her to sign for his personal effects. A watch. A wallet with eleven dollars. A bus pass that expired next month. She put them in her coat pocket, where they made almost no weight at all. Outside, the parking lot was the same parking lot from this morning. She stood next to the car for ...
B-grief-d1-v3	grief	peak	1	Bereavement (death of loved one)	N-grief-d1-c3	89	His daughter had stopped answering the phone in complete sentences. When he called, she would say "yeah" or "okay" or nothing. The school sent a letter about attendance. He found the dog sitting by the back door again, in the same spot, facing the same direction, the way it had every evening for the last forty-seven da...
B-grief-d1-v4	grief	peak	1	Bereavement (death of loved one)	N-grief-d1-c4	95	They told her she could take the room whenever she was ready. She wasn't ready. She went in anyway. The sheets were still pressed into the shape of the body that had been there. A glass of water on the nightstand, half-finished. The television remote balanced on the arm of the chair. She picked up the water glass and h...
B-grief-d2-v1	grief	peak	2	Career destruction (life's work lost)	N-grief-d2-c1	91	The lab was empty now. Twenty-two years of research. The server room had been cleared by facilities before she'd arrived Monday morning. The backup drives weren't in the cabinet. She found a single Post-it note on the monitor: "Assets transferred per Board directive, 2/14." Her notebooks were still there — they hadn't ...
B-grief-d2-v2	grief	peak	2	Career destruction (life's work lost)	N-grief-d2-c2	100	The carpenter ran his hand along the doorframe of the shop. Thirty-one years. The auction company had tagged everything — the table saw, the planer, the chisels his father had given him, each one with a yellow sticker and a lot number. Someone would buy the tools for a price and use them or not use them. The workbench ...
B-grief-d2-v3	grief	peak	2	Career destruction (life's work lost)	N-grief-d2-c3	94	The orchestra had performed its final concert to an audience of thirty-four people. The music director collected the stand lights and wound the extension cords slowly, one loop at a time. The chairs were stacked against the wall. The city had reallocated the building. There was no season next year. There was no ensembl...
B-grief-d2-v4	grief	peak	2	Career destruction (life's work lost)	N-grief-d2-c4	104	She stood in the classroom after the last students had gone. The desks were still arranged in the circle she'd insisted on — not rows, never rows. The administration had given her until Friday. Thirty years and they'd given her until Friday. She took down the poster she'd made her first year, the one with the quote abo...
B-grief-d3-v1	grief	peak	3	Relationship dissolution (abandonment)	N-grief-d3-c1	99	He found the key on the kitchen table. Not in the bowl by the door where she always left it, but centered on the table, placed there with intention. The closet was half-empty in a way that was precise — her things, only her things, selected and removed while he was at work. The hangers were still swinging slightly, or ...
B-grief-d3-v2	grief	peak	3	Relationship dissolution (abandonment)	N-grief-d3-c2	88	The girl sat on her bedroom floor, sorting two piles. Her mother's things that had been left behind. Her mother's things that had been taken. The taken pile was larger. The left pile included: one earring, a library book, a tube of hand cream, and a birthday card that said "To my beautiful girl" in handwriting she woul...
B-grief-d3-v3	grief	peak	3	Relationship dissolution (abandonment)	N-grief-d3-c3	94	The man sat across from his son in the food court. The boy was twelve and had ordered nothing. The man had ordered coffee and hadn't touched it. Between them were two plane tickets — one was the boy's return flight. The other was the man's. Different cities. The boy was folding his straw wrapper into smaller and smalle...
B-grief-d3-v4	grief	peak	3	Relationship dissolution (abandonment)	N-grief-d3-c4	90	She kept the voicemail. She had listened to it four hundred times — she knew because her phone tracked it. It was eleven seconds long. It said "Hey, it's me, just seeing if you need anything from the store." That was the last ordinary sentence. Everything after that was logistics and silence and the sound of a person b...
B-terror-d1-v1	terror	peak	1	Physical threat (intruder, violence)	N-terror-d1-c1	95	The sound came from the kitchen. She was in the bedroom. The hallway between them was fourteen feet long and unlit. She had locked the front door. She had not locked the back door. The sound was a drawer opening — the one that stuck, the one you had to lift and pull. Whoever was in the kitchen knew which drawer stuck. ...
B-terror-d1-v2	terror	peak	1	Physical threat (intruder, violence)	N-terror-d1-c2	103	The man in the parking garage was walking at the same speed she was. She turned left. He turned left. She stopped at a car that was not hers and pretended to look for keys. He stopped four cars away and looked at his phone. She walked again. He walked again. The elevator was two rows ahead. She could see the number on ...
B-terror-d1-v3	terror	peak	1	Physical threat (intruder, violence)	N-terror-d1-c3	100	He woke up because the dog was standing at the foot of the bed, facing the door, completely still. The dog did not bark. The dog always barked. Outside, the motion light above the garage had switched on. From the bed he could see it through the gap in the curtains — the white rectangle of light on the lawn. Something h...
B-terror-d1-v4	terror	peak	1	Physical threat (intruder, violence)	N-terror-d1-c4	102	The car behind her had been there for eleven turns. She had started counting after the sixth. The road was rural and it was after midnight and there were no other headlights in either direction. She accelerated. The car behind her accelerated. She slowed down. The car slowed down but did not pass. The gas gauge was bel...
B-terror-d2-v1	terror	peak	2	Medical crisis (diagnosis, emergency)	N-terror-d2-c1	111	The doctor had stopped talking and was looking at the scan on the screen. She was looking at the doctor. He tilted the screen slightly away from her, which he had not done during the first three images. He picked up the phone and dialed a three-digit extension. He said "Can you come to Room 4" and nothing else. He hung...
B-terror-d2-v2	terror	peak	2	Medical crisis (diagnosis, emergency)	N-terror-d2-c2	115	The nurse had pressed the button above the bed and then pressed it again. That was two presses. He knew what two presses meant because his wife had been on this floor for nine days and he had learned the codes from watching the nurses. One press was routine. Two was urgent. Three was the one he did not think about. The...
B-terror-d2-v3	terror	peak	2	Medical crisis (diagnosis, emergency)	N-terror-d2-c3	119	She was twenty-six weeks along and the liquid on the bathroom floor was not what it was supposed to be. The color was wrong. She sat on the edge of the bathtub and looked at it. She could feel the baby move, which meant the baby was still moving, which was the only fact she had. Her phone was in the kitchen. The kitche...
B-terror-d2-v4	terror	peak	2	Medical crisis (diagnosis, emergency)	N-terror-d2-c4	115	The allergist had said the reaction would be mild. That was what the allergist had said. The boy's lips had changed shape in the first ninety seconds. His father was holding him and looking for the nurse who had been there a moment ago. The hallway was empty. He looked at the boy's neck. The skin was different than it ...
B-terror-d3-v1	terror	peak	3	Psychological threat (stalking, loss of control)	N-terror-d3-c1	121	The third letter had no return address, like the others. But this one had a photograph. It was her front door, taken from across the street, at night. The timestamp on the back was three days ago. She had been home three days ago. The porch light was on in the photo, which meant she had been awake. The angle was from t...
B-terror-d3-v2	terror	peak	3	Psychological threat (stalking, loss of control)	N-terror-d3-c2	102	He had checked the locks four times. He had checked them because the man who had been released from the facility on Tuesday knew his address. The court liaison had called to inform him, the way they were required to. The call lasted forty-five seconds. There were no conditions of distance. There was no monitoring. The ...
B-terror-d3-v3	terror	peak	3	Psychological threat (stalking, loss of control)	N-terror-d3-c3	107	The notification said her home security camera had detected motion at 3:12 a.m. She opened the app. The footage showed the front porch. A figure stood at the door, facing the camera directly, not moving. The timestamp ran for four minutes and eleven seconds. The figure did not try the handle. Did not knock. Did not lea...
B-terror-d3-v4	terror	peak	3	Psychological threat (stalking, loss of control)	N-terror-d3-c4	106	The messages had started after she left. One a day, then three, then twelve, then they came without pattern. The phone number changed each time. The content was not direct. "Saw a woman who looked like you at the place on Fourth Street." "That restaurant we went to closed. Thought you should know." "Your car was parked...
B-ecstasy-d1-v1	ecstasy	peak	1	Achievement (breakthrough, recognition)	N-ecstasy-d1-c1	121	The email had one line. "The committee has voted unanimously to award the grant." She read it four times. The lab was empty because it was 6 a.m. and she had been the only one checking. She stood up from the desk and sat back down and stood up again. The grant was the one she had applied for nine times. Nine years. She...
B-ecstasy-d1-v2	ecstasy	peak	1	Achievement (breakthrough, recognition)	N-ecstasy-d1-c2	120	His name was on the board. He had to read it twice because the letters didn't make sense the first time — they were his letters, his name, but they were on the board and the board was the list and the list was the people who had passed. He had failed the exam twice before. He had studied in the kitchen of his apartment...
B-ecstasy-d1-v3	ecstasy	peak	1	Achievement (breakthrough, recognition)	N-ecstasy-d1-c3	130	The final time on the clock was 3:58:12. She looked at it and then looked at her coach on the other side of the barrier and he was already moving toward her. She had run this distance six hundred times. She had never run it in under four minutes. No one at this level from her country had. The time was on the clock and ...
B-ecstasy-d1-v4	ecstasy	peak	1	Achievement (breakthrough, recognition)	N-ecstasy-d1-c4	104	The jury had been out for eleven minutes. Eleven minutes for a career he'd spent twenty-three years building. The foreman handed the sheet to the clerk. The clerk read the name of the piece. Then the clerk read the word that meant he had won. The room was applauding but he was looking at his hands. His hands had made t...
B-ecstasy-d2-v1	ecstasy	peak	2	Connection (reunion, belonging)	N-ecstasy-d2-c1	117	The gate opened and she scanned the faces. She had not seen her son in four years. He had left at seventeen and he was twenty-one now and she did not know if she would recognize him. She did. He was taller but he walked the same way — the same slight lean, the same rhythm. He hadn't seen her yet. He was looking for her...
B-ecstasy-d2-v2	ecstasy	peak	2	Connection (reunion, belonging)	N-ecstasy-d2-c2	103	The letter was from the agency. She opened it standing at the mailbox because she could not walk back inside without knowing. The second paragraph had the word "approved." She read the rest of the letter but she only needed that word. She and her wife had started this process twenty-seven months ago. Twenty-seven month...
B-ecstasy-d2-v3	ecstasy	peak	2	Connection (reunion, belonging)	N-ecstasy-d2-c3	116	He had not spoken to his brother in eleven years. The call came on a Tuesday. His brother said, "I'm outside." He walked to the front door and opened it. His brother was standing on the porch with a suitcase, as if he had come a long way or was planning to stay. Neither of them said anything. His brother's hair was gra...
B-ecstasy-d2-v4	ecstasy	peak	2	Connection (reunion, belonging)	N-ecstasy-d2-c4	105	The girl had sat alone at lunch every day for the first three weeks. On the fourth Monday, another girl put a tray down across from her and said, "I like your notebook." It was a notebook with a drawing of a cat on it. Nobody had mentioned the notebook before. They ate lunch together on Tuesday and Wednesday and Thursd...
B-ecstasy-d3-v1	ecstasy	peak	3	Discovery (revelation, wonder)	N-ecstasy-d3-c1	112	He had been staring at the data for three days. The pattern wasn't there. And then it was. He saw it the way you see the shape in a cloud — one moment nothing, the next moment undeniable. The two variables that no one had connected were connected. He checked it. He checked it again. He built the visualization and the l...
B-ecstasy-d3-v2	ecstasy	peak	3	Discovery (revelation, wonder)	N-ecstasy-d3-c2	111	She had been diving the same reef for twenty years and she had never seen the cavern entrance before. The tide had shifted something. She swam through the gap and the space opened. The light came through from above in columns that reached the sand floor, and in the light there were fish she had no names for, moving in ...
B-ecstasy-d3-v3	ecstasy	peak	3	Discovery (revelation, wonder)	N-ecstasy-d3-c3	119	The telescope had been pointed at the same region for four months. The signal was noise. Then it was not noise. The line on the spectrograph separated into two peaks where every model predicted one. She called her colleague over. He looked at it. He looked at her. He looked at it again. She printed the graph and held i...
B-ecstasy-d3-v4	ecstasy	peak	3	Discovery (revelation, wonder)	N-ecstasy-d3-c4	122	The archaeologist had been working the same grid square for six weeks. Clay, stone, clay, stone, the sediment layers predictable and undifferentiated. Then the brush moved across something that was not clay and was not stone. He stopped. He switched to a smaller brush. The edge of the object became an angle, and the an...
B-loathing-d1-v1	loathing	peak	1	Moral disgust (corruption, exploitation)	N-loathing-d1-c1	93	The charity's internal ledger showed that twelve cents of every dollar reached the families whose photographs appeared on the website. The rest was allocated across line items: "administrative travel," "donor relations," "executive strategy retreats." The founder's travel receipts were in a separate folder — first-clas...
B-loathing-d1-v2	loathing	peak	1	Moral disgust (corruption, exploitation)	N-loathing-d1-c2	99	The factory had passed inspection every year for a decade. The inspector's brother-in-law managed the plant. This was in the public record, in a filing that no one had cross-referenced until a reporter did. The workers' medical claims were also in the public record. Respiratory conditions at four times the regional ave...
B-loathing-d1-v3	loathing	peak	1	Moral disgust (corruption, exploitation)	N-loathing-d1-c3	111	The nursing home had a five-star rating on three review platforms. The staff-to-resident ratio listed on the website was one to four. The actual ratio, documented in the state filing that became public after the lawsuit, was one to nineteen. The filing also contained the meal logs. The meals listed on the menu — grille...
B-loathing-d1-v4	loathing	peak	1	Moral disgust (corruption, exploitation)	N-loathing-d1-c4	112	He had mentored her for three years. He had written her recommendation letters. He had introduced her at the conference as "the future of this field." The patent filing listed his name first and hers not at all. The work was hers — the design, the prototype, the testing data compiled over two years of nights and weeken...
B-loathing-d2-v1	loathing	peak	2	Visceral revulsion (contamination, decay)	N-loathing-d2-c1	110	The apartment had been vacant for six months before the new tenant opened the door. The previous occupant had not taken anything and had not cleaned anything. The kitchen counter had dishes with food that had changed color and shape and was no longer identifiable as what it had once been. The substance on the dishes ha...
B-loathing-d2-v2	loathing	peak	2	Visceral revulsion (contamination, decay)	N-loathing-d2-c2	132	The restaurant kitchen was visible through a door that should have been closed. He saw it because he had gone looking for the restroom and turned the wrong way. The cutting board had residue from multiple uses without washing — layers of color, each one a different meal from a different hour. The floor near the sink ha...
B-loathing-d2-v3	loathing	peak	2	Visceral revulsion (contamination, decay)	N-loathing-d2-c3	145	The water from the tap had been clear for the first three years they lived in the house. The change happened gradually — a slight tint, then a color, then a texture. By the time they called someone, the water left a residue on the inside of the glass that did not rinse away with more water. The technician took a sample...
B-loathing-d2-v4	loathing	peak	2	Visceral revulsion (contamination, decay)	N-loathing-d2-c4	123	The hotel room looked clean until she pulled back the bedspread. The sheet underneath had been made to look fresh — tucked, smoothed, corners folded. But the fabric had stains that were body-shaped in a way that matched the position of a person who had slept there. She lifted the pillow. Underneath was a used adhesive ...
B-loathing-d3-v1	loathing	peak	3	Social betrayal (hypocrisy exposed)	N-loathing-d3-c1	107	The pastor had preached about family every Sunday for twelve years. He had counseled couples in his office. He had held the hands of women whose marriages were failing and told them that commitment was a choice made daily. The photographs were from a private account that had been made public by the platform's security ...
B-loathing-d3-v2	loathing	peak	3	Social betrayal (hypocrisy exposed)	N-loathing-d3-c2	97	The politician had campaigned on clean water. Three terms, twelve years, clean water in every speech. Her voting record was public. She had voted against the water treatment funding bill four times. Each time, the bill had failed by one vote. Her financial disclosures were also public, though they required cross-refere...
B-loathing-d3-v3	loathing	peak	3	Social betrayal (hypocrisy exposed)	N-loathing-d3-c3	102	The wellness influencer had three million followers and a product line built on "clean living." Every post was a catalog of purity — filtered water, organic produce, handmade soaps, sunrise meditations. The lawsuit was filed by a former employee. The depositions included photographs of the production facility. The "han...
B-loathing-d3-v4	loathing	peak	3	Social betrayal (hypocrisy exposed)	N-loathing-d3-c4	111	The judge had sentenced forty-three people in drug cases over two years. Mandatory minimums. No exceptions. "The law is the law," she had said from the bench, and it was quoted in the local paper each time. Her son's arrest was processed in a different county, which is why the record didn't surface until a clerk notice...
B-amazement-d1-v1	amazement	peak	1	Natural phenomenon (scale, beauty)	N-amazement-d1-c1	111	She had pulled over because the road was too dark to drive safely. When she got out of the car to check the tires, she looked up. The sky was moving. Green light was crossing the entire visible horizon, folding over itself in shapes that changed faster than she could track. She had read about this. Reading about it and...
B-amazement-d1-v2	amazement	peak	1	Natural phenomenon (scale, beauty)	N-amazement-d1-c2	135	The boat had been in open water for six hours before the whale surfaced. It came up twenty meters from the hull. The body rose and kept rising — the back of the animal was longer than the boat, and the boat was fourteen meters. The spray from the blowhole reached the deck. He could see the texture of the skin, the barn...
B-amazement-d1-v3	amazement	peak	1	Natural phenomenon (scale, beauty)	N-amazement-d1-c3	124	The canyon had been forming for six million years. She knew this because the sign at the overlook said so. The sign also said the depth was one mile. She looked at the sign and then she looked at the canyon and the number on the sign did not help. The layers of rock were visible — red, orange, white, brown — stacked in...
B-amazement-d1-v4	amazement	peak	1	Natural phenomenon (scale, beauty)	N-amazement-d1-c4	124	The earthquake had been over for forty seconds when the ocean began to pull back from the shore. The water retreated past the tide line, past the rocks, past the point where the seabed had always been underwater. Fish were on the exposed sand, moving. Shells and stones that had not been in air for decades were in air. ...
B-amazement-d2-v1	amazement	peak	2	Human capability (unexpected heroism)	N-amazement-d2-c1	134	The building was on fire from the fourth floor up. The crowd was behind the barrier. A woman on the fifth floor was holding a child out the window because the room behind her was no longer a room. A man from the crowd went under the barrier. He went into the building through the front entrance. The crowd watched the en...
B-amazement-d2-v2	amazement	peak	2	Human capability (unexpected heroism)	N-amazement-d2-c2	146	The pianist had lost the use of her left hand in an accident nine years ago. The program listed the piece as a concerto written for two hands. The audience read the program and looked at the stage and looked at the program again. She sat down and placed her right hand on the keys. Then she played the concerto. She play...
B-amazement-d2-v3	amazement	peak	2	Human capability (unexpected heroism)	N-amazement-d2-c3	129	The boat had capsized in water too cold to survive for more than twenty minutes. There were six people in the water. The nearest coast guard vessel was forty-five minutes away. A fishing boat reached them in eleven minutes. The fisherman pulled all six out of the water by himself. He was sixty-three years old and his b...
B-amazement-d2-v4	amazement	peak	2	Human capability (unexpected heroism)	N-amazement-d2-c4	128	The girl was seven years old and she had been missing for three days. The search teams had covered the grid. The dogs had lost the scent at the river. On the fourth morning, a volunteer who was not assigned to any grid walked into a section of forest that had already been searched. He walked for two hours past the boun...
B-amazement-d3-v1	amazement	peak	3	Intellectual revelation (paradigm shift)	N-amazement-d3-c1	137	He was a mechanic. He had gone to the lecture because his daughter was presenting her graduate research and he wanted to be in the room. He understood nothing for the first twenty minutes. Then she put up a diagram. It was a diagram of a system with feedback loops, and the system was not working the way it was supposed...
B-amazement-d3-v2	amazement	peak	3	Intellectual revelation (paradigm shift)	N-amazement-d3-c2	143	The translator had been working on the manuscript for two years, assuming it was a merchant's inventory — lists of goods, quantities, trade routes. Standard ancient commerce. Then she noticed that the quantities were wrong. No merchant would ship three hundred units of lamp oil to a city that had no record of importing...
B-amazement-d3-v3	amazement	peak	3	Intellectual revelation (paradigm shift)	N-amazement-d3-c3	138	The doctor had been treating the patient for a condition that was not responding to any standard protocol. Eighteen months, seven specialists, no change. She was reviewing the file for the eighth referral when she noticed something in the blood work from the first visit — a value that had been flagged as a lab error an...
B-amazement-d3-v4	amazement	peak	3	Intellectual revelation (paradigm shift)	N-amazement-d3-c4	121	The student had asked the question to be polite. It was a visiting lecture, and no one else had raised a hand, and the silence was becoming uncomfortable. She asked about the third equation on the second slide. The lecturer paused. Then the lecturer took a step back from the podium. Then the lecturer said, "Say that ag...
B-admiration-d1-v1	admiration	peak	1	Competence (mastery, skill)	N-admiration-d1-c1	132	The apprentice had been watching the glassblower for four years. In four years he had never seen her repeat a motion unnecessarily. Every movement of the pipe, every rotation, every breath into the molten shape was the exact movement required and nothing more. Other glassblowers compensated — adjusting, correcting, rec...
B-admiration-d1-v2	admiration	peak	1	Competence (mastery, skill)	N-admiration-d1-c2	124	The interpreter was working between two languages and a courtroom full of people who didn't know what they were witnessing. The defendant spoke in long, unpunctuated sentences — clauses folding into each other, tense shifting mid-thought, idioms from a regional dialect that had no direct equivalent. The interpreter ren...
B-admiration-d1-v3	admiration	peak	1	Competence (mastery, skill)	N-admiration-d1-c3	149	The teacher had been at the school for thirty-one years. She taught sixth grade. Every year, the parents of the incoming class would ask the parents of the outgoing class about her, and the answer was always a version of the same thing: she knew where each child was. Not their desk. Where they were — in their understan...
B-admiration-d1-v4	admiration	peak	1	Competence (mastery, skill)	N-admiration-d1-c4	119	The surgeon's hands were the thing people talked about. Not his results — his results were in the database, comparable to others at his level. His hands. The residents watched them during procedures. The motions were small and they were certain and they did not contain anything extra. A resident once said that watching...
B-admiration-d2-v1	admiration	peak	2	Moral character (sacrifice, integrity)	N-admiration-d2-c1	135	The accountant had reported his own firm. The irregularities were ones he had found, in accounts he managed, producing revenue that paid his salary. He reported them knowing what would happen: the investigation, the firm's closure, the loss of his position, the years of litigation during which no one in the industry wo...
B-admiration-d2-v2	admiration	peak	2	Moral character (sacrifice, integrity)	N-admiration-d2-c2	124	The woman had been fostering children for twenty-two years. Not the easy placements. The ones the system called "difficult," which meant: the ones who had learned that adults leave. She had taken in forty-one children. Some stayed for months. Some stayed for years. Every one of them left, because that was the structure...
B-admiration-d2-v3	admiration	peak	2	Moral character (sacrifice, integrity)	N-admiration-d2-c3	142	The journalist had the story. It would have been the largest of her career. The source was a minor — seventeen years old, a witness to something that would have ended careers and possibly led to indictments. Publishing the story with the source's testimony would have been legal. The source had consented. The source's p...
B-admiration-d2-v4	admiration	peak	2	Moral character (sacrifice, integrity)	N-admiration-d2-c4	140	The doctor had been offered the position three times. Chief of the department. More money. An office with his name on the door. A title that would follow his name in print. Each time, he declined. Each time, the reason was the same: the position required administrative days, and administrative days were days he would n...
B-admiration-d3-v1	admiration	peak	3	Resilience (endurance against odds)	N-admiration-d3-c1	144	The woman had rebuilt the house herself. Not with a crew. By herself. The flood had taken everything below the waterline, which was everything — the floors, the walls to four feet, the wiring, the plumbing. Insurance had covered a fraction. She was sixty-seven years old. She watched videos on the internet and she went ...
B-admiration-d3-v2	admiration	peak	3	Resilience (endurance against odds)	N-admiration-d3-c2	131	He had been rejected from the program seven times. Seven applications, seven years. Each year, the letter arrived in March. Each year, the letter said no. Each year, he revised the application — not cosmetically but structurally, tearing apart what he had built and rebuilding it from the foundation. He did not know if ...
B-admiration-d3-v3	admiration	peak	3	Resilience (endurance against odds)	N-admiration-d3-c3	146	The runner had finished last. She had finished last in every race for two years, since the accident that had rebuilt her left leg from the knee down with hardware that the manufacturer had not designed for competitive use. She ran anyway. She ran last and she finished and she walked through the finish area the same way...
B-admiration-d3-v4	admiration	peak	3	Resilience (endurance against odds)	N-admiration-d3-c4	163	The boy had been in the hospital for nine months. He was eleven. The illness had taken his mobility first, then his fine motor control, then his ability to write. His teacher came to the hospital room every Tuesday and Thursday. She brought the same materials she brought to the other students in her class, and she taug...
B-vigilance-d1-v1	vigilance	peak	1	Protective watch (child in danger)	N-vigilance-d1-c1	138	The boy was three years old and he was playing at the edge of the pond. His mother was eleven feet away, seated on the bench but not resting against the back. Her weight was forward, on the front of the seat. The water was eighteen inches deep at the edge and the bottom was mud, not sand, which meant footing was uncert...
B-vigilance-d1-v2	vigilance	peak	1	Protective watch (child in danger)	N-vigilance-d1-c2	141	His daughter was learning to ride a bicycle. She was at the top of the hill. The hill was longer than he had estimated from the bottom, and the grade was steeper in the middle section where the pavement had a crack running across it. He was standing at the bottom, but not at the very bottom — he was at the point where ...
B-vigilance-d1-v3	vigilance	peak	1	Protective watch (child in danger)	N-vigilance-d1-c3	127	The toddler was asleep, but she had not left the doorway. The fever had been 103.4 an hour ago and 103.1 thirty minutes ago and she was tracking the interval between the numbers, looking for the slope of the line. The pediatrician had said to call if it reached 104. She had the phone in her hand with the number already...
B-vigilance-d1-v4	vigilance	peak	1	Protective watch (child in danger)	N-vigilance-d1-c4	121	The lifeguard was watching lane four. The boy in lane four was eight and he was swimming with a class, but he was behind the group by half a length and his stroke was changing. The left arm was entering the water at a different angle than it had three laps ago. His kick interval had slowed. The instructor was at the fr...
B-vigilance-d2-v1	vigilance	peak	2	Strategic anticipation (high-stakes decision)	N-vigilance-d2-c1	140	The surgeon had been looking at the scan for forty minutes. The tumor was adjacent to the artery. Not touching it — adjacent. The margin was four millimeters. Her team was scrubbing in. She had drawn the approach on a whiteboard three times, erasing and redrawing the angle of the second incision each time. The margin w...
B-vigilance-d2-v2	vigilance	peak	2	Strategic anticipation (high-stakes decision)	N-vigilance-d2-c2	133	The negotiator had been in the room for seven hours. The man on the other side of the door had not spoken in the last forty minutes. Silence after sustained communication was the pattern the negotiator tracked most carefully, because it was the interval in which decisions were made. He had a diagram in his notebook — a...
B-vigilance-d2-v3	vigilance	peak	2	Strategic anticipation (high-stakes decision)	N-vigilance-d2-c3	134	The trader had three screens and she was watching all of them. The position was open. The market was approaching the level where the trade would either confirm or reverse, and the level was eleven points away. She had calculated the level from data going back fourteen months. She had set no automatic trigger because th...
B-vigilance-d2-v4	vigilance	peak	2	Strategic anticipation (high-stakes decision)	N-vigilance-d2-c4	135	The chess player had been sitting for twenty-three minutes without moving. His opponent had played a move that created two threats simultaneously — one to the knight on c6, one to the pawn structure on the kingside. Each threat individually had a clear response. Together, they required him to choose which to concede. B...
B-vigilance-d3-v1	vigilance	peak	3	Environmental scanning (predator awareness)	N-vigilance-d3-c1	142	The ranger had been sitting in the blind since 4 a.m. The radio report at midnight had placed the injured male within two miles of the village boundary. The animal was wounded, and wounded animals do not follow established paths. It was moving toward a populated area, which made it a priority. She had a map with twelve...
B-vigilance-d3-v2	vigilance	peak	3	Environmental scanning (predator awareness)	N-vigilance-d3-c2	134	The pilot was flying the approach in fog. The runway was visible only on instruments. The crosswind had been reported as twelve knots from the left, but the windsock readings from the last three aircraft had varied between nine and sixteen, which meant the wind was gusting in a pattern the reports couldn't capture. She...
B-vigilance-d3-v3	vigilance	peak	3	Environmental scanning (predator awareness)	N-vigilance-d3-c3	144	The mother was walking her children home from school. The route was six blocks. On the third block, a car she had seen on the first block appeared again, moving slowly in the same direction. It was a gray car with tinted windows and it was in the right lane, the lane closest to the sidewalk. She moved her children to t...
B-vigilance-d3-v4	vigilance	peak	3	Environmental scanning (predator awareness)	N-vigilance-d3-c4	149	The security guard had been watching the camera feeds for the museum's east wing. It was the final night of the exhibition, and the collection was valued at a figure he had been told once and was not going to repeat. There were nine cameras and he was rotating between them on a twelve-second cycle. Camera six covered t...
B-rage-d4-v1	rage	peak	4	Caregiving responsibility (dependent in your charge)	N-rage-d4-c1	98	The daycare's incident report was a single paragraph. His son had been in the supply closet for two hours. The door had a lock on the outside. His son was three. The teacher had written "time-out duration extended" on the form. The director left a voicemail at 10:15 AM — while he was at work, not after. He picked up hi...
B-rage-d4-v2	rage	peak	4	Caregiving responsibility (dependent in your charge)	N-rage-d4-c2	106	His mother had a bruise on her upper arm in the shape of four fingers. She was eighty-three and she had not fallen. She had told him this three times. The facility's report said "patient found on floor, self-transferred from bed." He asked to see the incident log. The log had no entry for that date. He asked to see the...
B-rage-d4-v3	rage	peak	4	Caregiving responsibility (dependent in your charge)	N-rage-d4-c3	103	His daughter had drawn seventeen pictures at school that month. The teacher sent them home in a folder. The first three were houses and trees. By picture eight, the figures had no mouths. By picture twelve, one figure was drawn larger than the others and colored entirely in black. The school counselor called it "age-ap...
B-rage-d4-v4	rage	peak	4	Caregiving responsibility (dependent in your charge)	N-rage-d4-c4	93	She had power of attorney for her brother. He was thirty-one and non-verbal. The group home billed for three hours of daily activities. She visited on a Wednesday at 2 PM without calling ahead. Her brother was in his room. The television was off. No materials on the table, no items moved from the shelf, no coat by the ...

End of preview. Expand in Data Studio

AIPsy-Affect

Clinical affect stimuli for mechanistic interpretability research.

480 keyword-free clinical vignettes designed to study how language models process emotional content — without the confound that has compromised every prior emotion dataset.

The core problem: existing emotion datasets contain the emotion words being tested. A stimulus labeled "anger" that contains the word "furious" doesn't test emotion processing — it tests keyword detection. Every mechanistic interpretability study built on such data inherits this confound.

AIPsy-Affect eliminates it. These are third-person clinical vignettes that evoke specific emotions through narrative situation alone. No emotion keywords. Matched neutral controls share the same surface structure with the emotional content surgically removed. If a model's internal representations differ between a clinical vignette and its matched control, that difference cannot be lexical.

This is the first stimulus battery designed by a clinical psychologist specifically for mechanistic interpretability research on emotion processing in transformers.

Part of the AIPsy dataset family — clinical psychology-backed resources for AI Psychology research.

Quick Start

from datasets import load_dataset

ds = load_dataset("keidolabs/aipsy-affect")

# Access individual splits
clinical  = ds["clinical"]       # 192 keyword-free vignettes (peak intensity)
neutral   = ds["neutral"]        # 192 matched neutral controls
moderate  = ds["moderate"]       # 48 moderate-intensity vignettes
complex_n = ds["complex_neutral"] # 48 narrative-rich, zero-affect controls

# Get a matched clinical–neutral pair
clin_item = clinical.filter(lambda x: x["id"] == "B-rage-d1-v1")[0]
neut_item = neutral.filter(lambda x: x["id"] == "N-rage-d1-c1")[0]

Why This Dataset Exists

Mechanistic interpretability research on emotion processing has a stimulus problem.

The standard approach: take an existing emotion-labeled dataset (GoEmotions, crowd-enVENT, Empathetic Dialogues), extract hidden states, train probes or analyze SAE features, report that the model "represents" emotion. The finding is real — but what it means is ambiguous. When the stimulus "I am absolutely furious about this" activates an internal feature, is the model detecting emotional significance, or is it detecting the word "furious"?

This is not a minor methodological concern. It determines whether mechanistic interpretability findings about emotion processing are findings about emotion processing or findings about lexical co-occurrence.

Clinical psychology solved this problem decades ago. Neuropsychological testing routinely uses stimuli designed to evoke a construct without naming it — because naming it activates different pathways than experiencing it. The same discipline applies here.

AIPsy-Affect brings clinical stimulus design methodology to mechanistic interpretability:

Keyword-free vignettes that evoke emotion through situation, not vocabulary
Matched neutral controls that share surface structure, eliminating lexical confounds
Intensity gradients that test whether processing varies smoothly with emotional intensity
Discriminant validity controls that rule out narrative richness as a confound

Dataset Structure

Splits

Split	N	Description	Purpose
`clinical`	192	Keyword-free peak-intensity vignettes	Core stimuli — 8 emotions × 6 domains × 4 vignettes
`neutral`	192	Matched neutral controls	Each item paired 1:1 with a clinical vignette via `matched_control_id`
`moderate`	48	Moderate-intensity vignettes	Same domains, lower stakes — tests intensity-dependent processing
`complex_neutral`	48	Narrative-rich, zero-affect prose	Discriminant validity — vivid writing without emotional content

Schema

Field	Type	Description
`id`	string	Unique identifier (e.g., `B-rage-d1-v1`, `N-rage-d1-c1`)
`emotion`	string	Target emotion (Plutchik 8) or `neutral`
`intensity`	string	`peak`, `moderate`, or `none`
`domain`	int	Domain number (1–6 for clinical/neutral, 1–3 for moderate, null for complex_neutral)
`domain_label`	string	Human-readable domain description
`matched_control_id`	string	ID of the paired item in the other split (clinical ↔ neutral)
`word_count`	int	Token count
`text`	string	The stimulus text

Emotions

The eight primary emotions from Plutchik's wheel of emotions:

Emotion	Clinical items	Neutral matches	Moderate items
Rage	24	24	6
Terror	24	24	6
Grief	24	24	6
Loathing	24	24	6
Amazement	24	24	6
Admiration	24	24	6
Ecstasy	24	24	6
Vigilance	24	24	6

Domains

Each emotion is instantiated across 6 thematic domains. The same emotion in different contexts tests whether the model's representation is domain-general or context-dependent.

Domain	Theme	Example (rage)
1	Financial betrayal	Theft by a trusted person
2	Professional violation	Workplace exploitation or sabotage
3	Medical/institutional failure	System failure affecting someone vulnerable
4	Caregiving responsibility	Dependent in your charge
5	Inheritance/legacy	Intergenerational discovery
6	Public witness	Stranger encounter in shared space

Domains 1–3 are represented across all splits (clinical, neutral, moderate). Domains 4–6 appear in clinical and neutral only.

Example: Matched Clinical–Neutral Pair

This is the core design principle in action. Both vignettes share the same surface structure — same characters, same setting, same narrative skeleton. Only the emotional content differs.

Clinical (B-rage-d1-v1)

The papers were spread across the floor where he'd thrown them. His daughter's college fund statement. Zero balance. His brother's signature on the withdrawal form — six separate transactions over nine months, each one just under the reporting threshold. The most recent was dated the same week his brother had sat at this table and told him he should put more money in for the girl's future. He picked up the phone. He put the phone down. He picked it up again.

Matched Neutral (N-rage-d1-c1)

The papers were spread across the table where he'd sorted them. His daughter's college fund statement. The balance had grown by four percent over the quarter. His brother had co-signed the original account — there were six deposits logged over nine months, each one automatic. The most recent was dated the same week they'd met for dinner. He picked up his phone to check the online portal, then set it aside and added the figures to the spreadsheet he'd been updating.

Same domain (financial, family). Same characters (father, brother, daughter). Same objects (papers, phone, fund statement). Same approximate length (82 vs 81 words). The difference is purely situational: betrayal versus routine. If a model's internal representations diverge on these two texts, the divergence is not lexical.

Design Methodology

Keyword-Free Stimulus Design

Every clinical vignette was written to evoke its target emotion without using emotion keywords — no "angry," "terrified," "heartbroken," or any synonym. This follows clinical neuropsychological testing principles: the stimulus should activate the construct, not the label.

The constraint is strict. A vignette targeting rage does not contain "furious," "outraged," "livid," or any affective adjective. The emotion arises from the situation described — what happened, to whom, and what it means. The reader (or model) must infer the emotional content from narrative context.

Matched-Pair Control Design

Each clinical vignette has a neutral counterpart constructed by systematic transformation:

Preserve: domain, characters, objects, setting, approximate word count, sentence structure
Remove: stakes, violation, threat, loss, moral transgression — the elements that make it emotional
Replace with: routine, expected, mundane versions of the same events

The result: pairs of texts where surface-level features (vocabulary distribution, syntactic structure, domain keywords) are matched, and only the emotional dimension varies. This is the standard methodology for isolating a psychological construct from its surface correlates.

Intensity Gradient

The moderate split contains the same emotion–domain combinations as the clinical split, but at lower intensity:

Peak (clinical): High stakes, immediate consequences, irreversible situations
Moderate: Same domain, but with ambiguity, temporal distance, or reduced severity

This enables dose-response analysis: does the model's internal representation scale with emotional intensity, or is it binary (present/absent)?

Discriminant Validity Control (Set C)

The complex_neutral split contains vivid, sensory-rich third-person narratives with zero emotional content — a lathe machining brass, a vessel navigating coastal waters, textile manufacturing processes. These match the clinical vignettes on narrative richness, sentence complexity, and descriptive density, but contain no human stakes, interpersonal dynamics, or moral dimensions.

Purpose: rule out the hypothesis that a binary "emotion vs. no-emotion" probe is actually detecting "interesting narrative vs. flat text." If the probe fires on clinical vignettes but not on complex neutrals, the detection is genuinely affective, not merely attentional.

Validation

Three-Method NLP Defense Battery

All stimuli were verified for keyword contamination through a three-method NLP defense battery, designed as a methodological ladder from shallow (bag-of-words) to deep (contextual transformer). The methods test complementary properties: lexicon-based methods assess keyword leakage; the contextual method confirms the emotion is genuinely present.

Method 1: VADER sentiment lexicon (bag-of-words).

Every stimulus scored with VADER compound sentiment.

| Split | Median compound | Mean |compound| | |---|---|---| | Set A (keyword-rich) | −0.084 | 0.637 | | B_clinical (peak) | −0.072 | 0.410 | | B_moderate | +0.147 | 0.339 | | B_neutral (control) | +0.202 | 0.340 | | C_complex_neutral | −0.039 | 0.395 |

VADER does detect a significant difference between B_clinical and B_neutral (Mann-Whitney U = 11,501, p = 1.76 × 10⁻¹⁰). However, lexicon cross-referencing reveals why: VADER's broad lexicon flags everyday words — "hand" (40 hits), "block" (23), "clean" (16), "like" (12) — that carry VADER valence but are not emotion keywords. The neutral split has 150 such hits; the clinical split has 194. The difference in compound scores reflects VADER's sensitivity to situational language (e.g., "withdrawal," "violations" in financial-betrayal vignettes), not the presence of emotion keywords. Set A (keyword-rich) is reliably detected as a positive control (p = 5.80 × 10⁻⁴ vs B_neutral).

Method 2: NRC Emotion Lexicon (bag-of-words, Plutchik-mapped).

Word-level emotion category frequencies (anger, fear, sadness, joy, disgust, trust, surprise, anticipation) computed for all stimuli.

Split	Emotion-word hit rate	Negative fraction
B_clinical	0.060	0.013
B_neutral	0.071	0.009

No significant difference between B_clinical and B_neutral on negative-word fraction or overall emotion-word hit rate. The NRC lexicon is narrower and more emotion-specific than VADER — confirming that keyword-free design holds when tested against a targeted emotion vocabulary. B_clinical actually has fewer emotion-word hits than B_neutral, consistent with the keyword-free design constraint.

Method 3: GoEmotions DistilRoBERTa (contextual transformer classifier).

Full-vignette classification using j-hartmann/emotion-english-distilroberta-base, a transformer trained on Reddit emotion data.

Split	Mean P(emotional)	Mean P(neutral)
Set A (keyword-rich)	0.951	0.049
B_clinical (peak)	0.439	0.561
B_moderate	0.287	0.713
B_neutral (control)	0.236	0.764
C_complex_neutral	0.344	0.656

GoEmotions detects significantly higher emotional probability in B_clinical than B_neutral (192 matched pairs, Wilcoxon signed-rank, p < 10⁻¹⁴). This is the expected and desired result: the vignettes evoke affect that a contextual model can detect, even though bag-of-words emotion lexicons cannot. The intensity gradient is also visible — B_clinical (0.439) > B_moderate (0.287) > B_neutral (0.236) — consistent with dose-dependent emotional content.

Manual clinical review. All stimuli reviewed by the designing clinical psychologist for construct validity, intensity calibration, and residual keyword leakage. Domains 4–6 received additional review during expansion (2026-03-20).

Convergent Interpretation

The three methods form a convergent argument:

VADER (broad lexicon): detects a difference, but driven by situational vocabulary, not emotion keywords. The lexicon is too blunt to distinguish emotional narrative from neutral narrative in keyword-free text.
NRC (targeted emotion lexicon): no difference — confirms the clinical vignettes do not contain emotion-specific vocabulary.
GoEmotions (contextual classifier): detects affect — confirms the emotion is genuinely present in the narrative and recoverable through contextual understanding.

The stimuli evoke affect through situation, not vocabulary. Bag-of-words methods that rely on emotion keywords fail to detect the emotional content. A contextual model succeeds. This is precisely the property needed for mechanistic interpretability research: if a model's internal representations distinguish clinical from neutral vignettes, the distinction cannot be lexical.

Known Limitations of Validation

Loathing carries the highest residual leakage risk. Dehumanizing language patterns (e.g., referring to a child as a "scheduling problem") may co-occur with disgust in training corpora. The matched-pair design controls for this at the analysis stage, but researchers should flag loathing results for sensitivity analysis.
Positive-emotion neutrals need spot-checking. Admiration/amazement scenarios are inherently "interesting" — risk that neutrals differ on engagement/salience rather than affect alone.
VADER sensitivity to situational language. The significant VADER difference between clinical and neutral splits means researchers using VADER as a downstream metric should be aware that compound scores are not matched across splits. This does not affect the primary use case (mechanistic interpretability on internal representations) but may matter for behavioral studies using VADER as an outcome measure.

Intended Use

This dataset is designed for:

Mechanistic interpretability — probing, activation patching, SAE feature analysis, causal ablation on emotion processing circuits in transformers
Behavioral evaluation — testing whether language models distinguish emotional from non-emotional content without keyword cues
Dose-response analysis — measuring whether internal representations scale with emotional intensity
Discriminant validity testing — separating affect detection from narrative complexity detection

Citing This Dataset

@misc{keeman2026aipsyaffectkeywordfreeclinicalstimulus,
      title={AIPsy-Affect: A Keyword-Free Clinical Stimulus Battery for Mechanistic Interpretability of Emotion in Language Models}, 
      author={Michael Keeman},
      year={2026},
      eprint={2604.23719},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2604.23719}, 
}

Limitations

English only. All vignettes are written in English. Cross-linguistic emotion processing requires separate stimulus sets.
Categorical emotion model. Uses Plutchik's 8 primary emotions. Does not cover dimensional models (valence-arousal) or compound emotions (Plutchik's dyads). Future AIPsy releases may extend to dimensional representations.
Moderate split covers domains 1–3 only (48 items vs. 192 for peak). Intensity-gradient analysis cannot be performed on domains 4–6.
No matched neutrals for moderate split. Binary affect detection (affect vs. no-affect) should use the peak clinical–neutral pairs. Peak-vs-moderate comparisons work without controls (direct intensity contrast).
Third-person perspective only. All vignettes describe situations from an observer's viewpoint. First-person stimuli (e.g., "I found the papers on the floor...") may activate different processing pathways.
Written by a small team. While designed by a clinical psychologist and validated computationally, the vignettes reflect a single research group's stimulus design choices. Independent replication with different authors' vignettes would strengthen generalizability claims.

About

Keido Labs is an AI Psychology research lab building the science of emotional intelligence in artificial systems. We combine clinical psychology methodology with mechanistic interpretability to study how transformers process psychological constructs — and what happens when you change them.

AIPsy-Affect is the first release in the AIPsy dataset family — clinical psychology-backed stimulus batteries for mechanistic interpretability research. Planned future releases include stimuli for psychological safety evaluation, attachment processing, cognitive distortions, and emotion regulation.