The Dartmouth Prison Experiment
I left my PhD program in Dartmouth’s Department of Psychological and Brain Sciences last November. And while it’s true that that Department is not a good place to be in general, my egress was due specifically to the toxic conditions within the Cognitive and Affective Neuroscience Lab, led by Tor Wager.
When I spend intensive time learning one subject, I find that as I grow an intuition for the lay of the land, I begin to forget the details that originally led to my understanding. What I’ve written here is a painfully brief tableau of many months of notes I took to avoid forgetting the details of how the CANLab is a poisonous place to be. A pithy (read: 9-page) description felt preferable to an enumeration of inhumanities or a decision tree produced from the many hours of processing I did on the phone with my lifelines outside of Dartmouth.
My purpose in writing this is fourfold:
First, if you or a loved one is considering pursuing a PhD in the Dartmouth PBS department, I want you to have this type of information that is hard to find on the website. Many students in the department share the sentiment of “if we get lots of good people here the culture will get better!”, and they are happy to talk admiringly about a place they personally cannot wait to leave. I find it problematic to mislead incipient scientists about what they will find once they move to the Middle-of-Nowhere, NH. Whether or not it is worth it to join the program is up for debate, but this honest description of my negative experience is at least now available.
Second, I hope this piece will give second thoughts to those considering joining the CANLab. I know how easy it is to watch the fMRI Methods coursera and the online lectures, and maybe even meet the wizard in person, and think “this is the lab where I’ll answer the most interesting neuroscience questions of all time.” I totally feel that. I was right there with you. I drank the kool-aid and went for it. I left just before my third year began because of the way Tor treats people under his regime.
Third, and this is the longest shot of all, I hope this will in some way provoke Tor to pause and spend meaningful time ensuring that his staff, the people who keep him famous, are treated with humanity. I don’t think he’ll ever read this, but maybe somehow the message will get passed along.
Fourth, writing this is therapeutic to me personally. I’ve been told that even within my own lab, the message people received about my departure primarily included vague explanations that I was unhappy. Of course I was generally unhappy. But other than the interaction with Tor I describe below, I was personally treated quite well. Watching the incessant and unmitigated neglect of those with the least power in the lab is what ultimately tipped the scale. It still often troubles me, while I’m falling asleep at night, that some people in the department have no idea why I left. It troubles me that I didn’t do more to help the workers I left behind. I won’t pretend that pacifying these demons and their ilk isn’t part of this writing. But I will say it ranked 4th.
I don’t expect this account to be taken with grace by anyone with power at Dartmouth, so I’ve necessarily omitted some very sad details for fear of any retaliation against those still in the lab I left.
And with that, I’ll start at the end. Below is the message I sent to Tor the day I quit. I then explicate some of the references I made in the letter. I end with a few caveats necessary for such a complex social situation and a final lamentation.
My Resignation from PBS
Dear Tor,
I am writing to give formal notice that I will be leaving the Dartmouth PBS program on Friday, November 26th.
This was a very difficult decision to make. After our meeting on September 23rd I’ve spent significant time reflecting on our advisor-advisee relationship and the way others in the lab are treated. I believe there is no longer a place for me in the CAN Lab. I’m writing this to you because after that fateful interaction I don’t feel comfortable bringing up private issues with you in person: one of the main reasons I feel I cannot successfully remain in the program.
I want to be clear that when I began the process of completing my master’s degree this year, I was not intending to leave the PhD program. But being yelled at and cursed at changed everything for me. I began searching for evidence of support from you. Instead, I found myself looking back to when you told me it was a risk to accept me as a graduate student and to when you dismissed my hybrid hyperalignment work as irrelevant. A few weeks later you skipped my specialist presentation. The next week you shortened my master’s defense calendar event less than an hour before I was scheduled to present. The truth is that I don’t feel supported by you as a scientist, a student, or an employee.
Tor, I want you to know that there are people in your lab past and present who think you treat your workers and advisees poorly. I have been acutely aware of this since you yelled obscenities at me in a one-on-one meeting. But I have also known about the poor working conditions of research assistants in the lab for a long time, having brought the issue to your attention on multiple occasions. I no longer wish to be a part of a lab capable of making people miserable in the name of accomplishing science. I have therefore decided to forgo a career in neuroscience, something I’ve dreamed of for the better part of a decade, and officially end my work in the lab next Friday.
With all of this said, I don’t expect you would prefer I stay in the lab beyond the 26th. I will work on transferring all data and documentation I have to Discovery and returning all lab and department resources to [the lab manager] before I depart. I earnestly hope that you will think on what I have written here and contemplate the immense power you hold over your lab members.
Sincerely, Luke
SpaceTop: Where the Hours are Out of This World
“I no longer wish to be a part of a lab capable of making people miserable in the name of accomplishing science.”
Many of you in the field will soon be hearing about the “Spacetop” or “Spatial Topology” dataset collected by the CANLab. If IRB approval were required to protect the treatment and welfare of human research assistants, this study would have been terminated more than a year ago.
I learned about Spacetop, an R01, when I was a research assistant with the lab in Boulder. In the final months before the lab moved from CU Boulder to Dartmouth, the project came alive. It is a massive project in which nearly 200 participants were to endure four one-and-a-half-hour fMRI scans. During which they would watch lots of movies (haphazardly picked by me, a know-nothing research assistant (RA) at the time). Participants also undergo a variety of tasks that involve listening to and reading stories and receiving thermal pain. Sounds interesting, right? I think so! If an extraordinary amount of time and focus had been given to this project, if there had been a timeline and a quality project manager from the outset, if all of the things a high-quality research project require had been done, it would have ranked among the most interesting data objects ever collected. None of this was done. And mere neglect of an R01 was only the beginning.
I remember having meetings in Boulder where people like me, who had never designed or implemented an fMRI task, were assigned to do just that. And in a hurry. Additionally, postdocs leaving the lab as it moved from Boulder were designing tasks, though it was clear they were not invested beyond the next few months. The reason untrained and uninterested people were slated to design the study was because the NIH, unlike the CANLab, does enforce deadlines, and by the time the project was being discussed it was already very behind.
We threw together a series of slipshod tasks, some of which had no clear hypotheses. Those would of course be figured out retrospectively, once that sweet sweet data was in hand. The plan was to collect as many participants as possible in Boulder, and then quickly transfer the entire protocol to Dartmouth. So in the last few months before the big move, something like 25 participants were technically run. I say “technically” because in almost every session, one of our hastily written tasks broke mid-run and had to be skipped. We found out later that much, if not all, of the data collected in Boulder was useless because while the fMRI data exists, the bug-ridden tasks were not collecting things like timestamps – data that would allow us to know what the participant was seeing and doing as their brains were being scanned. In fact, I hope all of the data is useless, because if it isn’t, some unsuspecting research assistant will one day be assigned the fool’s errand of trying to salvage something meaningful from the flotsam. The point here is that the project was behind schedule and poorly managed from the outset.
And then things began to deteriorate. Once at Dartmouth, a graduate student in the lab was put in charge of Spacetop. Finally, someone on the ground with fMRI know-how and an eye for analysis to manage the details and make the decisions necessary to drive the project forward. This person would work with the research assistants (who actually set up and run everything) while interfacing with Tor to give project updates and get expert opinion where needed.
The issue is that the project was already insurmountably behind schedule. Why? Mismanagement. And it was made clear to us, over and over, exactly how far behind schedule it was. “We need to collect 120 participants in a year.” This is equivalent to sixty hours of scanning each month. But scans take time to set up and tear down, and every scan included a 30-90 minute behavioral session beforehand. Participants also needed to be recruited, scheduled, and rescheduled. The faulty equipment used for pain stimuli needed to be maintained and often sent for repairs. The list goes on. All of this, of course, was to be managed by the 3 research assistants with ‘guidance’ from the project manager. Then, just as we had begun debugging the problematic code from Boulder and setting up the tasks in the Dartmouth scanner suite, the COVID-19 pandemic began.
I’ll tell you now: Spacetop is currently on track to very nearly meet its original (pre-COVID) data collection quota. Pause for a second and ponder. How could a project, already months behind on scanning head into a maelstrom of disease, social distancing, scanner shut-downs, and new study restrictions, and come out the other side ostensibly on track to meet a goal that was set with no knowledge of those obstacles?
The answer is that its workers were abused.
Please appreciate the context. The Spacetop team consisted of the graduate-student-project-manager, three research assistants, and me. All three research assistants had aspirations of applying to graduate school. This means they needed to secure strong letters of recommendation from Tor. They had to impress him. But Tor isn’t impressed by anyone doing one huge job really well. So these RAs needed to work at breakneck speed on Spacetop, while also finding other projects within the lab to convey the kind of simulacrum of breadth of which Tor himself is a master. One RA was also literally the Lab Manager of the entire lab. This alone set the stage for these brilliant young scientists to work long, stressful hours. And they did. All the while their work environment continued to sour.
The project manager, having been in another lab at CU Boulder, was also somewhat new to the CANLab and was even more eager to impress the boss. She needed this project to move and move fast. Having written multiple tasks for the study, she could see, with dollar signs gleaming in her eyes, the careeristic potential of being the graduate student to usher in this massive Wager-branded neuroscientific product. So she pushed. And pushed. And pushed. If Tor said “jump”, she would have said “show me whose head to stand on.”
I facetiously titled this writing “The Dartmouth Prison Experiment,” but a more accurate title is probably “The Dartmouth Milgram Experiment.” The Milgram experiment was a social psychology classic in which participants believed they were delivering increasingly harmful electric shocks to another person. However, the other people were not actually being shocked at all. An authority figure would instruct them to crank up the intensity of the shocks to levels the participants believed to be fatal. And a very high proportion of the participants would continue to follow the instructions of the authority figure. In Spacetop, Tor was/is the authority figure calling out the demands. “More data, faster, we’re behind.” And the project manager was the Milgram participant, shocking the research assistants with chronic stress, long hours, and uncertain futures. Unfortunately, in the Spacetop variation of the experiment, the pain and suffering was real.
We were rushing, and haste makes waste. So a fear-and-blame-based management style was employed. When something new inevitably went wrong, the person responsible was criticized in front of the team either in-person or on Slack. If something went wrong in someone’s presence, instead of asking for help they were encouraged to point out who was at fault – lest they be blamed. When an RA successfully completed a task, they were handed a never ending to-do list. They were working 80+ hour weeks, constantly at the lab setting up Spacetop. When they went home, they would work on side projects to keep up the appearance of breadth required to impress Tor. Afraid of being castigated, or worse, reported to Tor for laziness, people on the team were canceling plans with loved ones and skipping doctor’s appointments. Taking a full weekend off was out of the question.
Since the project manager was the only line of communication up to Tor, the RAs didn’t know what he was being made aware of. Consider how stressful that uncertainty can be. Did he know the RAs were working double their hours (i.e. working for half their contractual rate)? Did he know they were skipping medical procedures to mask-up and scan participants late into the night, putting themselves at risk of contracting COVID-19? Or was he only being made aware of their shortcomings? We had reason to believe the latter because team discussions were essentially blamestorming sessions. Tor didn’t attend our team meetings, so the RAs would try to schedule short check-ins with him. Which he often failed to show up for (without notice). As a graduate student I was only peripherally affected by this caustic existence, and for the first time in my life I was losing sleep because of work. I can only imagine what it would have felt like if my livelihood were on the line.
Time-to-time the issue was raised that the Spacetop RAs should be compensated for their work that required more than the 40 hours per week set in their contracts. Each time the project manager would explain that she brought it up with Tor, and that it was simply not going to happen. The RAs were also strafed repeatedly by the project manager with the following piece of academic mentorship handed down to her from Tor: “No successful neuroscientist has ever worked only 40 hours a week.” As if this divine academic axiom was license to abuse.
The message to the RAs was clear. Work hard. Work long. Don’t expect facetime. Don’t expect remuneration. Will all of this get you into grad school? Maybe, have you taught yourself fMRI analysis and completed a project in your spare time?
You don’t have to be a Wagerian brain genius to know that chronic stress of this sort is unhealthy. One day an RA messaged me to come down to the scanner. They were running a participant and the task broke (over a year into the project). That same task had already broken during an earlier scan that day, and the project manager had had to come down to fix it the first time. The RA was terrified of having to ask for help from the project manager a second time. Afraid they would be upbraided. Afraid that they would be perceived as incompetent. Afraid that Tor would hear about their perceived incompetency. So they fearfully reached out to me instead. While I was fixing the code (which broke through no fault of this RA’s) they paced nervously back and forth in the scan suite. Fidgeting with their hands and explaining how scared they felt. This is a workplace. When the scan was up and running, they threw their arms around me in relief. Should people feel the need to hug a coworker for receiving a little help? I was nonplussed.
Closer to my quitting the lab, an RA was chatting with me while running a participant:
“Luke, I hate this lab. I’m miserable and stressed constantly. I can’t sleep. But I don’t have any other options. Where would I go?”
In my final months in the lab I brought these things up to Tor on multiple occasions. The first time he seemed sincerely troubled by the news. That was nice, but also consider how aloof one would have to be to remain oblivious to such a horrific environment in their own lab. He said he would talk with the project manager about moving away from a fear-based style of authority. He also copied and pasted a Slack message to each of the RAs to say he appreciated all their hard work. This was a meretricious concession that cost him no time. About a month later (after nothing changed) I brought the issue to his attention a second time and was again met with superficial shock and empty gestures. To my knowledge, the Spacetop work environment today remains as toxic as ever.
A final cherry on top of this hellworld sundae is the consideration of authorship. I spent many hours trying to convince the project manager (another grad student) that our RAs should be guaranteed authorship on at least the first wave of papers to come from the massive dataset they had created for her and Tor through sheer moxie and misery. But she insisted that authorship on a paper requires “an intellectual contribution.” In my opinion, the magnitude of intellectual onanism required to use a phrase like “intellectual contribution”, given the context, boggles the mind.
Without exploiting these scientists’ labor your publications would never exist. Plain and simple. You owe them your career (plus over time). Silently add their names to the author list, tweet out your publication link, and continue acting like you’ve done something admirable. You lose nothing, they gain one more miniscule advantage. This is, how does one say? A no brainer.
I must admit to my own cowardice here. I brought up these issues to Tor multiple times, but I made no demands. I should have said “Pay your workers over time when they do your science outside of work hours. Do not ask them to work a single hour over 40 each week. Acknowledge their exceptional performance and write them gushing letters of recommendation today. Personally introduce them to the researchers they want to study under now. Give them the time off they’re owed. Do all of this in the next four weeks or I quit.” But I didn’t. I submitted and defended my master’s degree, wrote Tor one last email, and left Dartmouth forever.
I hope future researchers who make use of the Spacetop data will know the conditions that led to its creation.
Working on the Railroad
“I’m writing this to you because after that fateful interaction I don’t feel comfortable bringing up private issues with you in person: one of the main reasons I feel I cannot successfully remain in the program.”
The most surreal experience I had in graduate school was being screamed at by the man I followed across the country to learn from. Months before resolving to quit my program, I decided I wanted to use the published paper I had co-first-authored as my master’s thesis. Working alone on a PhD is a difficult thing to do in a pandemic. Especially when your only interpersonal interactions are by way of a poisonous work environment. I had hoped the degree would help me make meaning of my work thus far, crystallize my accomplishments, and galvanize my effort heading into my third year. Instead it would reveal to me how little Tor valued me as his student.
I reached out to the department’s director of graduate studies to ask if using my paper for the thesis would be an acceptable thing to do. He had no problem with it and suggested I form my committee and discuss it with them. The paperwork to formalize my committee required Tor’s approval. Because Tor is spread so thin, I sent the document to him via Slack and email, with an email bump the next day. Usually it takes him a week or more to read a message, so this was standard operating procedure.
He replied to my email saying, “Using a paper you led and first-authored would be OK in principle. In this case, this is a paper you assisted with but didn’t first author.” This is patently false, so I set up a meeting to discuss it. After canceling the meeting a few minutes before it was set to start, Tor later called me and explained a vague reluctance to allow a co-first-authored paper as a thesis. He continually expressed that he wanted to be sure he was consistent with departmental standards, while also acknowledging that the department had no standard for this (At Dartmouth, PhD students don’t automatically obtain master’s degrees while remaining on the PhD track.) After our call, Tor texted me saying “please check on the precedent and standards with [the heads of the department]. If it meets Dartmouth’s standards and your committee members agree then I will support it.”
In individual emails, the department’s director of graduate studies and my other committee members expressed no concerns (and in some cases, enthusiasm) over using my paper as a master’s thesis. I forwarded these responses to Tor, since this was explicitly my homework assignment, and asked him again to sign my committee paperwork. Once more I was met with reluctance, so I waited to speak with him in our next scheduled meeting.
When I sat down in his office I asked him directly why it felt like he was against me getting the master’s degree. And the wind changed direction and Tor began talking. And he kept talking, fomenting his own anger. My calm and sincere rejoinders were met with increasingly hostile answers and accusations.
There’s a look you may recognize, usually found in the eyes of a child who has just realized they’ve been slighted in some way. It’s the look of a temper tantrum climaxing. Of compassion leaving the body and anger filling the empty space. And I saw that look in Tor’s eyes as he stood up now, commensurately raising the volume of his voice, and pointing at me. He exclaimed that he felt I was “railroading him” into making a decision about my master’s thesis. “It’s fucking wrong, Luke! It’s fucking wrong!”. Then he repeatedly yelled to me, with all the braggadocio of a famous academic ego, that he had published 10 papers during graduate school, asking rhetorically if that meant he should receive 10 masters degrees. As if my actions had somehow suggested that published papers and master’s degrees were infinitely fungible. He asked over and over if the other co-first author on my paper should receive a master’s degree from Dartmouth as well. All this before finally resorting again to the notion that the department must have standards for what is acceptable to use for a thesis. He suggested that, while all of these faculty members told me in writing that they were okay with my use of the paper, that they actually may be against it.
Maybe I should have worked harder to decode the true meaning of responses like “I would think that the paper is fully sufficient for a master’s thesis. I hope others agree.” But alas. Maybe I lacked the determination it takes to be a successful neuroscientist.
At the end of the meeting I explained to Tor why I wanted the degree in the first place. That I was miserable with the Spacetop environment and trying to make meaning of a life in the middle of nowhere during a pandemic. He texted me afterward to talk and let me know that he would allow me to use the paper for my thesis. I guess it doesn’t meet Dartmouth standards. Unless you’re sad enough.
There is one last disappointing detail of this wild incident: Because I never expected to be yelled at when I went in for my meeting, I left Tor’s office door open. There’s a famous social psychologist whose office is across the hall from Tor’s, and we exchanged a glance in the middle of his diatribe. They had to have heard some fraction of what was taking place. I made a mental note to look for a message from this person in the following week. Some small gesture just to say “I saw you may have been screamed at by your advisor, is everything okay?” I never heard from them again.
When Tor replied to my resignation email, he addressed his little episode by saying “When I swore during our meeting, it was to simply express frustration, as I see you as an equal and partner, and one whom I can speak frankly with.” I’m sure he and Kent Kiehl get on Zoom together all the time and really have an apoplectic go at each other, but this is not how I expect coworkers or mentors to behave. “Baby, can’t we talk about it? You know I care about and support you. I swore at you and made you feel scared and tried to intimidate you because I CARE.”
- All Healthy Relationships
Caveats
Some people will read all this and think “this just sounds like academia in general, what’s the big deal?” This is repugnant for all the obvious reasons. But if you’re confused why stories of tenured faculty abusing their subordinates in the Dartmouth PBS department may be of special consideration, one blog will never be enough.
I don’t think Tor is evil. I think in the case of Spacetop he was deeply obtuse and negligent, which in this case is indistinguishable from abusive. There are worse PIs than him out there. This excuses nothing that I’ve written above.
I don’t think the grad student managing Spacetop is evil. I think she wants what he has, which is why anyone chooses an advisor. But a good person with power under a culture of neglect and abuse will do evil things.
I don’t think anyone should go to Dartmouth PBS on purpose. But there are good labs and good PIs there. Primarily the untenured faculty. There are also good students. I still don’t think the research there is worth the department existing as-is.
“Why didn’t you just lay all of this on the table for Tor and try to work through it together before you left?” This is a valid question and about 2.5 months of processing went into not doing this. The short answer is twofold. Once I made up my mind to leave, I was worried that Tor would prevent me from getting my master’s degree before departing. In cowardice I quietly got my degree approved and left. The other piece is that initiatives to improve lab culture in the CANLab are generally “supported” by Tor – meaning he wishes they would happen all of their own accord. He expects someone else in the lab to do the work. I brought some of these issues to his attention on multiple occasions, but I don’t believe the problems can be solved by anyone but the man with the power.
“But the methods at Dartmouth are so cool!” So is paid labor.
A Final Lamentation
One of the most excruciating aspects of studying in the Dartmouth PBS Department was watching people who should leave decide to stay. These are students who loathe the department, hate the school, and have personal adversaries within both the student body and the faculty. They know leaving is in their best interest, they make plans to get out, to move across the country and never look back. And they never do it. Watching this happen, over and over, like a predetermined, West-Worldian cycle of suffering, honestly scared the shit out of me. I didn’t want to end up this way.
At the same time, I sincerely meant it when I wrote to Tor that a career in neuroscience was something I had dreamed of for the better part of the last decade. I finished a fifth year of undergrad in 2014, in which I studied cognitive and computer science, and I remember leaving Rochester thinking these were the most interesting topics in the whole world. Every educational and professional decision I made from that point forward was designed to get me into the field of computational neuroscience. And it worked. I tasted the privilege of chasing my intellectual whimsies. So I understand the addiction. Wanting my name on every paper, wanting my ideas to be the best and most clever and most complicated. I wanted my data to be the biggest and most impressive. My career was blossoming in my advisor’s image.
Looking back now, it was all so embarrassingly childish. Such boy-king, Trumpian delusions of grandeur. Grow the brand! Bigger data! Fancier methods! The important questions fading to dull susurrations under the din of the Big Important Neuroscientist’s manifest destiny. Who are we helping? And more importantly, who are we hurting?
The core of the CANLab, plain and simple, is Empire. More data, more analysis, more papers, more grants, repeat. And the Emperor is not interested in how it happens. Understanding the conditions of his workers would take time. It would slow the wheel. Of course he wants them to be happy. But given the choice between spending another hour paying half-attention on zoom so he can get his name on the next big “collaboration”, or checking in with his workers to earnestly make sure those who run his lab are surviving?
“Can someone send me the zoom link? I can’t find it.”