THE STUDY GROUP
Chapter One: The Wrong Plane Seat
The flight from Atlanta to Boston was delayed forty seven minutes, long enough for Nadia Osei to sit at Gate C14 and catastrophise in full, methodical detail.
You don't have to do this, she told herself, not for the first time that summer. You could call Professor Anand. Tell her the funding fell through. Tell her you've reconsidered.
She wouldn't, of course. At twenty five, she had spent four years working toward this exact moment: the Harlow North Cognitive Architecture Lab at Whitmore University, one of three programmes in the country doing the kind of interdisciplinary computational neuroscience she actually wanted to spend her career on. She had written the research statement four times. She had revised the writing sample until the margins bled red ink. She had turned down a perfectly reasonable offer from Michigan and a genuinely flattering one from Stanford to be here, in Gate C14, with a duffel bag of books and three weeks of anxiety compressed into a single departure lounge.
You're excited, she corrected herself. The catastrophising feeling and the excited feeling are physiologically identical. You've read the paper.
She had, in fact, read the paper. She'd written a summary of it for the writing sample. The distinction between terror and anticipation was, neurologically speaking, a matter of interpretation. She chose to interpret this as anticipation, and her heart, unhelpfully, kept thudding like it disagreed.
The flight was called. She found her seat, 14F, window, paid for extra because she'd intended to sleep through the whole thing, and was wedging her jacket into the overhead bin when someone dropped into 14E.
"Sorry," he said, and he didn't sound sorry. He sounded distracted, like a man whose attention was running six processes at once and the apology was a low priority background task. Laptop bag. Coffee cup sweating onto his wrist. The particular rumpled quality of someone who had been moving through airports since six that morning and had stopped minding it.
She looked at him the way you look at a seatmate: quick, assessing, calculating how the armrest negotiation was going to go. He is unexpectedly attractive, she thought, which was not a useful observation in an airport, and which she filed away accordingly, the way she filed away most things that were not useful.
He was perhaps thirty five. Dark hair that needed a cut, or had recently had one that hadn't quite settled. Glasses with a thin wire frame he pushed up his nose with one knuckle as he opened his laptop. A dark green sweater over a collared shirt, the specific uniform of a man who worked in academia or something close enough to it, for whom comfort and a modest kind of presentability were the only sartorial requirements he recognised.
Stop noticing, she told herself. Read your paper.
She got out her paper, a review article on predictive coding frameworks that she had already read twice and was going to read a third time because the first two readings had been interrupted by catastrophising. They took off. He worked with the concentrated energy of a man in the middle of something that mattered, and she read and tried, unsuccessfully, not to notice that he smelled faintly of coffee and something warmer underneath it, cedar maybe, and that his hands looked like they had done physical work at some point in his life. Broad. Capable. A small pale scar across the left index finger that her eyes kept returning to for no reason she could defend.
You are catastrophising, she told herself, and also apparently constructing elaborate backstories for strangers based on their hands. Stop.
Somewhere over Connecticut, without looking up from his screen, he said, "Predictive coding or active inference?"
She looked at him.
"Sorry?"
"The Friston review." He nodded at her paper. "Active inference or predictive coding? He conflates them in places. The distinction matters for what you're actually trying to do."
She looked at her paper. She looked at him. He had, in fact, glanced up, for the first time since takeoff, and was watching her with the polite, genuine attention of a man who had asked a question because he wanted the answer.
"Predictive coding, for now," she said. "I'm starting a PhD in September. Computational neuroscience."
"Where?"
"Whitmore."
He made a sound that wasn't quite a response, more a particular sharpening of attention. "Good programme," he said. "The HN Lab?"
"Yes."
"You'll be working with Harlow and North?"
"And Calloway," she said. "It's an experimental tri advisor structure. I applied specifically for it."
He looked at his screen for a moment, something working behind his eyes. "What's your research area?"
"Predictive processing models of social cognition," she said. "Specifically how the predictive brain handles belief updating in social contexts, when observed behaviour contradicts your model of someone's internal states."
He was quiet.
"That's interesting," he said, and it carried the specific weight of something he meant rather than something polite. "The mismatch signal question."
"Yes, exactly." Something in her shifted, the particular warmth of talking about her work with someone who was actually following. "The question is whether social prediction error functions the same way as sensory prediction error, or whether it's a distinct computational process. The clinical implications are significant either way. Autism, social anxiety, certain psychotic spectrum conditions."
"Bayesian models of mind," he said.
"Karl Friston's work, yes. Though there are good critiques of the framework too. Whether it's genuinely explanatory or just descriptive."
"Do you have a position on that?"
She looked at him. Most people she'd described this to, relatives, friends, the well meaning non specialists, wore the specific expression of someone performing interest. He was actually asking.
"Tentatively," she said. "I think the descriptive versus explanatory distinction is less meaningful than the critics suggest. All good models are descriptive first. The question is whether they generate novel predictions."
He nodded slowly. "Does the predictive coding framework generate novel predictions for social cognition?"
"That's the dissertation," she said.
He smiled, the first real one, something that reorganised his whole face from merely handsome into something far more interesting. A low pull went through her chest, unexpected and entirely unprofessional, and she pretended not to feel it.
"Fair answer," he said. "I'm Elias, by the way."
"Nadia." She watched his mouth shape her name and felt it land somewhere under her ribs. "Are you in neuroscience?"
"Adjacent. Computational cognitive science. Decision theory."
"That's a good intersection," she said. "Bayesian decision theory and predictive processing."
"It is." He turned fully toward her then, one knee angled into the narrow space between the seats, close enough that she could feel the warmth coming off him, and they talked the way you talk sometimes on planes, when the artificial intimacy of proximity and altitude produces conversation that would never happen anywhere else. He was sharp and specific, pushing back when he disagreed, building on what interested him, asking questions that made her think rather than simply explain. At one point his knee brushed hers and neither of them moved away, and she noticed that far more than she should have.
Four hours, roughly, between Connecticut and Boston.
She didn't know exactly when she stopped thinking of him as the stranger in 14E and started thinking of him as Elias, as someone she was genuinely talking to, the way she talked to maybe a dozen people in the world. Somewhere over Rhode Island it happened, and by descent she was describing her undergraduate thesis and he was describing a paper he'd been trying to write for two years, and they had moved from polite seatmate conversation into something with a different gravity entirely. Twice their hands landed near each other on the shared armrest. Twice she felt the small, electric awareness of not pulling hers away first.
The plane bumped down onto Logan's runway. Around them the cabin came back to life, phones leaving airplane mode, overhead bins clattering open, bodies unfolding from six hours of stillness.
"Good luck with the programme," Elias said, gathering his laptop bag, close enough that she caught the coffee and cedar smell of him one more time and let herself have it. "The HN Lab is excellent."
"Thank you," she said. "Good luck with the paper."
He smiled, and then the aisle filled and he was ahead of her, and they walked out of the terminal into the separate streams of their separate lives. I will never see him again, she thought, and that is slightly disappointing, and I should stop thinking about it. Her pulse did not immediately cooperate.
She collected her bags. She took a taxi to the address she'd been given. She was going to be a PhD student at Whitmore University, in one of the best programmes in the country for what she wanted to do, with three of the leading researchers in her field as advisors.
You're excited, she reminded herself. This is the thing you chose.
* * *
Orientation was in the Linden Building's main seminar room, a room she had looked up online, though the photographs hadn't captured how much light it got in the morning, or the particular smell of dark wood and old books and decades of academic work laid one on top of the other. She arrived seven minutes early, which left her time to choose a seat (middle row, slightly left of centre, visible but not conspicuous) and to look over the afternoon's programme.
Welcome address. Department overview. Introduction to faculty advisors. Lab tours.
She had read the faculty profiles online. Dr. James Harrow, thirty five, one of the architects of the tri advisor structure, known for theoretical precision and a teaching style students described variously as demanding, transformative, and the reason I finished my dissertation. Dr. Finn Calloway, thirty eight, the relational warmth of the three, his students writing acknowledgements that were notably personal, the kind that suggested he'd done more than advise their methodology. And Dr. Elias North, thirty six, the most recently tenured of the three, whose research on Bayesian models of social decision making was the primary reason she had applied to this programme at all.
The room filled. Six other doctoral students, by her count, this cohort deliberately small, the tri advisor structure requiring more intensive supervision than standard programmes could offer. Kwame from Yale. Priyanka from Edinburgh. Two from MIT. A student from Tokyo whose work on cross cultural cognition she was genuinely looking forward to reading.
The door at the front opened. The first faculty member entered. That is Dr. Harrow, she thought, recognising him from his photograph.
The second. That is Dr. Calloway, taller than the photograph suggested.
The third.
The third walked in and surveyed the assembled students, and Nadia felt the precise, lurching sensation of a predictive model failing, the mismatch signal, the exact thing she intended to spend her career studying, because the third man was Elias.
Not Elias from the plane. Not a Elias.
Dr. Elias North. The Bayesian decision theorist. The reason she had applied to this programme. The man whose last three papers lived in her annotated bibliography, whose research questions intersected hers in a way that had made her rewrite her research statement four times until it was exactly right.
The man she had talked to for four hours on a plane, their knees touching, his hand almost brushing hers.
He looked up at exactly the wrong moment.
Not at the room in general. At her. The specific seat she had chosen, slightly left of centre, in the middle row.
She watched him understand what he was looking at. It happened fast: recognition, recalibration, the precise thing a predictive processing model would describe as the instant a prior failed to match incoming evidence and the system updated. His expression didn't change, she would learn he was very good at not changing it, but something underneath the stillness of his face did. Her pulse climbed and would not come back down.
Nadia, it said. The student from the plane.
She held very still.
Dr. Harrow was speaking: welcome, congratulations, an overview of programme structure. She absorbed maybe thirty percent of it. The rest of her processing capacity was occupied with the recalibration of her own model, which had filed Dr. Elias North, primary research interest advisor, and Elias from the plane, the man she'd talked to about dissertations and Friston and the mismatch signal, into two entirely separate compartments, and was now discovering they contained the same person, the same hands, the same low voice she could still feel in her sternum.
The prior was wrong, she thought. The evidence is updating the model. This is exactly what I'm studying.
This is absolutely not what I expected to study.
* * *
After the welcome address there was a break, coffee and tea in the corridor, the specific negotiation of seven strangers becoming seven colleagues. Kwame found her first, warm in the way of someone who had developed warmth as a deliberate strategy until the strategy became the thing itself. They talked about his research, social network formation, the topology of trust, and hers, and the real and interesting intersection between them.
She was mid sentence with Kwame when she became aware that someone had joined them.
"Nadia," said Dr. North. "This is a strange coincidence."
She looked at him. He was doing something careful with his face, the neutrality of a man choosing it on purpose.
"Dr. North," she said.
"You didn't mention Whitmore."
"You didn't ask," she said. "You mentioned the HN Lab was a good programme. I didn't know you were at Whitmore."
"I'm aware." He looked at her for a moment with the quality she remembered from the plane, the visible processing, the thinking you could almost watch happen behind his eyes. "This is going to be somewhat complicated."
"Because of the four hours on the plane?"
"Because of the four hours on the plane," he agreed, and something about the low, exact way he said it made her stomach drop pleasantly, inconveniently.
Kwame looked between them with the expression of someone arriving at the end of a conversation rather than the beginning.
"Should I," Kwame began.
"No," she said. "It's fine. Dr. North and I were seatmates on the flight from Atlanta."
"Ah," said Kwame, which covered a generous range of possible meanings.
"We talked about research," she said, which was true, and also insufficient.
Dr. North said, "I should let you meet your other advisors." He held her gaze a beat too long for professional comfort. "Dr. Harrow will want to schedule your first one on one this week. Dr. Calloway tends to find students at these things rather than waiting for them to come to him."
"Thank you," she said.
He held her gaze one more moment, longer than was strictly necessary, and she felt it everywhere it had no business reaching, and then he moved on to Kwame, and she stood with her tea thinking about predictive models and the failure of priors and the particular problem of knowing someone's hands before you were supposed to know anything about them at all.
* * *
Dr. Harrow found her at the edge of the room.
He was younger than she'd expected, the photographs had suggested formality, someone who occupied authority seriously, and in person that was still true, but there was something else underneath it, the kind of intelligence that made her feel assessed and found interesting in the same instant.
"Nadia Osei," he said.
"Dr. Harrow."
"James. In one on ones. In seminars, Dr. Harrow, the students are more comfortable with the distinction." He glanced at her tea. "Your undergraduate work on mentalising and prediction error. I read it three times before the admissions committee meeting."
"Three times?"
"The argument in chapter four. About the relationship between social cognition and interoception. I disagreed with part of it. I wanted to be sure I was disagreeing with what you'd actually written rather than my own misreading of it."
"And were you?" she said. "Disagreeing with what I'd actually written?"
Something moved at the corner of his mouth. "About sixty percent," he said. "The other forty percent has been interesting for me to sit with."
"What part did you disagree with?"
"The claim that interoceptive signals are constitutive of social prediction rather than merely correlated with it," he said. "The neuroimaging evidence is suggestive but not decisive."
"The lesion studies are."
"Some of them. The patient populations are heterogeneous."
"That's the dissertation," she said, and felt a strange small vertigo, because it was exactly what she'd said to Elias on the plane.
James Harrow looked at her and then did something unexpected: smiled, full and sudden, changing his whole controlled face into something warmer.
"Good," he said. "We'll schedule Thursday. Ten o'clock."
He moved on. She stood with her tea and thought: three advisors. Three very different kinds of intelligence. One of them I already know. This is going to be complicated. And then, honestly, to herself alone: this is also exactly the kind of complicated I find interesting.
Dr. Calloway found her last, near the end of the corridor gathering. Warm was the wrong word for him, or not sufficient; he had the quality of a man for whom warmth was not a strategy but a structural feature, the way some people had perfect pitch, simply a true thing about him. Broad shouldered, easy in a room the way people are when they've never had to negotiate for comfort in one.
"Finn Calloway," he said. "You don't have to call me Dr. Calloway unless you want to."
"Nadia."
"I know." He said it without the careful framing of a compliment, just information delivered plainly. "Your research statement is the best I've read in four years of this programme. The question about social prediction error is exactly the right question. The methodology you proposed is ambitious but achievable." He looked at her properly then. "You're going to do something real here."
He means it, she thought. No performance in it, no social lubrication, just a man telling her something he believed.
"Thank you," she said.
"The tri advisor structure is going to be interesting for you specifically," he said. "Three people with three different relationships to your research area. James will push your theoretical rigour. Eli," he paused, barely, "Elias will push the Bayesian framework. I'm going to make sure you don't disappear into the methodology."
"What does that mean?"
"Some students, deep in a dissertation, lose the thread of why they started. I try to keep the thread visible." His eyes were steady on hers. "You started because you care about the clinical implications."
"I did."
"I know. It's in the statement. Don't let us argue you out of it."
"Will you?" she said. "Argue me out of it?"
"James might try," he said cheerfully. "He argues everyone out of everything for the first year and then argues them back in from a stronger position."
"And Dr. North?"
Something brief and specific crossed Finn Calloway's face.
"Elias is careful with what he pushes," he said. "You'll have to watch for the places where he's not saying something."
She thought of the plane, four hours in which she'd felt, genuinely, that he was saying everything, that the thinking was visible, that she could follow every step of it.
"He seemed fairly direct to me," she said carefully.
Finn looked at her for a moment.
"You've met already?"
"We were on the same flight," she said. "We talked."
"Ah." His voice stayed even. "That's interesting." He looked down at his coffee. "Elias is direct about his ideas," he said. "He's less direct about other things. You'll figure out the distinction."
The room was thinning, students finding each other, faculty dispersing, the afternoon's lab tours starting in ten minutes. She had the specific sensation of having arrived somewhere that was going to be far more complicated than its photographs suggested, and far more alive because of it.
This is exactly what you wanted, she told herself.
She looked across the room. Elias North was talking to another faculty member, and he glanced up at exactly the right moment, or the wrong one, depending on how you counted.
Everything is a prediction, she thought, her pulse still not entirely her own. Everything is a prior that the world is about to revise.
I'm going to need a lot of new priors.
* * *