What kind of thing or activity is a trolley car thought experiment? A philosophical puzzle, a classroom exercise, a tool for theoretical development, a laboratory experiment, a distortion of real moral dilemmas, or a total waste of time? Curiously, despite its abominable simplicity, the experiment can be any one of these multiple things, depending on who you ask. Among anthropologists, there is typically disinterest or suspicion of trolley problems. “Obviously the trolley scenario is highly artificial,” writes Webb Keane in setting up his book Ethical Life as an empirical study. “The ethical problem is presented as a discrete event that requires a single decision and transpires within a brief time frame” (2016: 7). For my part, the trolley problem is an exercise I run in the classroom at the beginning of term as a foil for how my course will study ethics anthropologically, which means putting moral problems into context, thickening the description as far as possible, and pushing students in the direction of nonjudgment. “We need not decide what to do or what someone else ought to do in this class,” I often say. “Let's focus on observing the situation first.” When running a thought experiment, on the other hand, I push students in the direction of being willing to make a split-second decision. “Assume you know x, y, and z, and make a choice!”

In this short article, I respond to this collection's call to contend with the appeal of stylized ethical dilemmas, thinking with rather than against them. I will do so by treating the trolley problem as a “found object,” an object like other found objects that exhibit the features of simplicity and artifice. I will focus on the pragmatics of form and ask what may come from an openness to self-consciously using imperfect devices, while knowing they may be discarded once they have served their purpose, or an experiment has failed.

I start by providing a found object of my own, and I propose a wild and rather imperfect analogy, a comparison of two things that are hardly the same thing: the philosopher's thought experiment and the psychologist's probe.

Between 2016 and 2019, I conducted fieldwork in a research institute for family therapy in a major cosmopolitan city in China. Part of a larger social trend in China known as the “psycho-boom”—the recent explosion of popular interest in all things psychology—this institute specializes in systemic therapy and offers public education, training courses, and clinical services, and it is highly idiosyncratic for bringing together, on the one hand, well-established psychiatrists who have worked “within the system” for decades and relative newcomers to the practice of psychological counseling, on the other. Compared to other approaches in the psycho-boom landscape, systemic therapy is not as popular let alone as well understood, difficult as it is to learn. It is, however, exceptionally good to think with and directly relevant for the purpose of this article. Based on the premise that psychiatric symptoms express relational problems, the institute uses a clinical protocol, when assessing new cases, that bears some similarity to the philosopher's experiment. The protocol, like a thought experiment, aims to reveal and to clarify, and it does so by creating a situation in which participants are also observers. Problems that are otherwise too vexing to confront in daily life become objects for critical reflection, even if only for a session or two.

           ^{* * *}

The initial clinical encounter at the institute is an assessment that is divided into two parts.¹ In the first part, the “conflict discussion,” two parents are asked to discuss a topic they disagree on. Meanwhile, the “identified patient” (IP)—the child or the adolescent—sits across from their parents in a chair on the other side of the room. The IP's fingers are attached to sensors tracking their heart rate, body temperature, and skin conductance. Behind the IP sit one or two assisting therapists who watch graphic bars on a computer screen fluctuate. They are taking notes on when the IP experiences arousals, indicated by moving bars, and in connection to which pieces of dialogue. They will report the arousals to the senior therapist, who is sitting on the other side of a one-way mirror, along with a report of the baseline, the set of data taken before the conflict discussion started.

The second part of the protocol is called the “debriefing.” A senior therapist joins the family on the other side with a colleague, and they will engage in a conversation with the family about their life together. The purpose is to encourage the family to reflect on how their inner lives are determined by patterns of interaction and subsystems that have formed and hardened over time. For example, the anger and resentment a teenager harbors against one parent does not belong to her alone; she is overly enmeshed and feeling such feelings on behalf of another parent. Therapy is guided by systemic theory and its ideas about the imperceptible formation of patterns as systems, which can then become circuits for the run of problematic energies and emotions. By externalizing hidden patterns and conflicts, therapy aims to persuade the family to see a psychiatric condition from a different angle, creating hope for change.

When the family refuses to engage in a conversation with the therapist during the debriefing, when they try to downplay a problem that the therapist is interested in exploring, the arousals report is invoked. In fact, therapy usually starts with the arousals report, consisting of numbers that appear neutral. If an IP is asked if she felt anything during the conflict discussion and the girl says no, a therapist may say “you have had an arousal, yet you did not even know it” (Kuan 2020: 703), generating skepticism about any claim any family member tries to make moving forward. By the end of a session, if all goes well therapeutically speaking, the family will have gained some insight into how they affect one another, which is the first step in liberating the adolescent from the family “mud pie” (Kuan 2020: 711). It is the first step in a process that aims to differentiate the ethical subjecthood of the “identified patient,” who has failed to disentangle herself from her parents. Between the arousals report, the questioning of claims and beliefs, and a one-way mirror that stages the therapy room as a field of observation, family members become participants and observers of their own dynamics. Although the arousals report appears to give a report of objective biofeedback data, the purpose of the report is not scientific accuracy. The report is an instrument for facilitating the eventual development of a self-determined, rational subject, and it aims to shake up family routines by casting doubt on habits of perception. Therapy typically ends with the question: “Illness or life, which one do you choose?” As if the choice was so simple.

           ^{* * *}

A few months into my fieldwork, the institute invited a psychologist from Hong Kong to give a presentation on the technology its clinical protocol depends on.² On that day, trainees who had enrolled in a supervision program offered by the institute for that year were gathered to learn how to operate the technology and how to record the data.

At one point, a brief but tense round of question-and-answer occurred between the woman giving the introduction and one of the trainees, a doctoral student in psychiatry. Ruiyun, the trainee, raised suspicions over whether the heart rate was really being measured; she argued actual heart rate can only be measured close to the heart. Amy, the guest speaker, tried to defend the technology with various explanations, including the possibility that something had been lost in translation—because the machines were developed in the United States—and what is being measured is in fact “blood pressure volume.”

With the presenter having the last word, Ruiyun turned to her neighbor, declaring that her doctoral training taught her the importance of rigor, as if to defend her challenge to our guest for the day. Meanwhile, I was confused by Ruiyun's insistence on accuracy, as I had been developing a different understanding of the purpose of collecting biofeedback data. Did Ruiyun fail to recognize a change of context? Between collecting data for clinical use on the one hand and collecting data for research purposes on the other (i.e., biofeedback data on how parental conflict affects a child's bodily responses), there is a difference in purpose and aspiration. The inventor of the biofeedback protocol has in fact clarified in a publication that it has the potential to “speed up the clinical process” (Lee et al. 2010: 45).

I have made my own share of “mistakes” in trying to transpose learning from one context to another. There was a time when, during a group supervision meeting, I posed a challenge to the teacher by asking about factors beyond the family, namely, the competitive education system and the pressure it exerts on young people. In response my question, the teacher made a firm request to the entire group: “Please do not pull in other systems; we are not here to teach you how to solve all the world's problems.”

Was she being reductive? Well, yes. She wants to focus on proximate causes, namely fixed patterns in kin relations. Is this teacher ignorant of other causes? I think not. Reduction, in this context, is an act of simplification that creates and defines a field of action, lodging a new possibility in a set of problematic relations in which action does not appear to be possible. (In family therapy, the immediate problem is commonly a disorder or set of symptoms that remains unresponsive to psychiatric treatment.) Therapy removes the noise of extrafamilial relations to concentrate focus on what appears near-at-hand. In trying to think comparatively about reduction in thought experiments and reduction as I ethnographically observed it, I appreciate Joshua Greene's response to criticisms of trolley dilemmas as having no worth because they are hypothetical and unrealistic: “Trolleyologists study simplified, hypothetical dilemmas not because they've failed to understand that the real world is complicated and uncertain. . . . To get serious about practical normative ethics, it helps to put aside the more mundane forms of complexity” (2023: 163). Greene explains that trolley problems are not meant to be “stand-ins” for real-life emergencies (2023: 164), nor are they meant to serve as personality tests or predictors of behavior. A trolley dilemma, like the biofeedback protocol, creates its very own context. Greene calls the former a “specialized probe” (2023: 171) for the study of mental phenomena, like the weird sentence “Colorless green ideas sleep furiously” used by Noam Chomsky in his study of syntax (2023: 163). This sentence does not intend to describe. Like an optical illusion (2023: 163–164), the aim is to “elicit competing responses” (2023: 164), the observation of which can teach us something about ourselves, Greene argues. I would add that the effect of a good probe may be disproportionately larger than the effort required to present it, depending on the genius of the design. In therapy, a specialized probe could “speed up the clinical process” (Lee et al. 2010: 45).

           ^{* * *}

In an essay titled “Experiments in Thinking about Observed Ethnological Material,” Gregory Bateson expresses his fondness for tinkering. The essay starts with observing a certain inheritance that helped to define anthropologists of his generation: suspicion against the use of loose analogies, a reaction against Herbert Spencer's influence. Bateson himself did not reject the use of analogy, attributing his idiosyncrasy to a “vague mystical feeling” he picked up from his father that everything is interconnected ([1972] 2000: 74). We might say Bateson's theoretical work is a transubstantiation of this feeling; his “wild hunches” became “strict formulations” ([1972] 2000: 87). Having described how animal segmentation facilitated his understanding of Iatmul social organization by providing a diagram and problems that were more easily observable, he goes on to discuss his efforts in developing a concept for capturing the “feel of culture,” which he would use as an aid for a study of the relation between the observable and something which is not. Both the analogy and the abstraction—that is, the concept of “ethos”—are, he admits tongue-in-cheek, offenses. But, he argues, “considerable contributions to science can be made with very blunt and crooked concepts” and “wild analogies” ([1972] 2000: 84, 87). In contemporary anthropology, we work with analogies all the time, for example when engaging a reference that may help us mount an argument, despite vast differences in the contexts to which the analogous cases belong. For example, one might find an anthropologist drawing a parallel between residents at a home for the cognitively disabled in Britain to “unruly medieval saints” (McKearney 2018: 55)—making a wild association in order to say something precise.³

Acontextual and apolitical, trolley problems cause offense to ethnographic sensibility and to the anthropological project more generally. But in philosophy, there are detractors too. As Hallvard Lillehammer observes in his introduction to the recent volume The Trolley Problem (2023): “There are those who consider the problem a conduit to discovering the nature of morality. Yet there are also those who consider it an example of academic theorizing at its most pointless” (2023: 1). The legal scholar Barbara H. Fried (2012) is most pointed in her attack on trolley problems for their artificiality (they consist of a set of known harms rather than unknown risks), and for taking attention away from what is really at stake (developing a nonconsequentialist understanding of trade-offs in situations of scarcity). Meanwhile, “big data trolleyology” has managed to use an abstract hypothetical scenario to generate an enormous data set for the study of cross-cultural variation, representing 70,000 respondents from 42 countries (Awad et al. 2020). The number is staggering. Trolley problems are certainly generative of debate and activity.

While this article has compared the trolley problem with a clinical protocol for addressing adolescent distress, a rather imprecise and awkward comparison, I wish to end by presenting a thought experiment of my own. Please fill in the blank in following sentence: “In a certain museum of found objects in a nameless part of the world, we may find on display the philosopher's thought experiment, the psychologist's probe, and the anthropologist's _______.”