Volume 16, Number 2 (2026)
Special Volume
The 1986 Stanford Conversation between Quine, Davidson, Dreben, and Føllesdal
Edited by Benjamin Marschall, Cambridge University and Andreas Føllesdal, University of Oslo

Triangulating with an Envatted Brain

James Pearson james.pearson@bridgew.edu Bridgewater State University
Abstract

During a conversation about skepticism at a closed 1986 Stanford Conference, Donald Davidson remarks that an envatted brain might correctly be interpreted as talking about the world outside its vat, to which it could have access via an “indirect causal chain.” In this paper, I examine this puzzling suggestion in the context of Davidson’s theory of triangulation, which would seem to preclude the interpreter being a mediate cause in the chain stemming from objects to the subject of interpretation. By prying apart a range of scenarios that the conference participants discuss, I distinguish Davidson’s antiskeptical argument from others in the semantic externalist family to which it belongs, and argue that his approach to language and mind may provide particular consolation to a person made anxious by the threat of solipsism.

Without [the] sharing of reactions to common stimuli, thought and speech would have no particular content—that is, no content at all. It takes two points of view to give a location to the cause of a thought, and thus to define its content. We may think of it as a form of triangulation: each of two people is reacting differentially to sensory stimuli streaming in from a certain direction. Projecting the incoming lines outward, the common cause is at their intersection. (Davidson 1991b, 213)

Skepticism emerges as a major topic of debate in the complete transcript of the Stanford 1986 Conference that has recently been discovered, at which W. V. Quine, Donald Davidson, Burton Dreben, and Dagfinn Føllesdal discussed various problems salient to Quine’s and Davidson’s philosophy. Davidson, in particular, tries to convince the others that his account of triangulation yields an antiskeptical argument that is unavailable to Quine. In the course of clarifying this proposal the participants discuss a range of scenarios, many of which involve the by now infamous figure of an envatted brain that is being manipulated by advanced scientists. If a coherent story can be told in which such a brain is deceived into believing that it perceives the external world, then what assurance can there be that we are not in a similar predicament, or that any of our beliefs about the world are accurate? The answer, Davidson thinks, lies in understanding the triangular structure that obtains between the interpreter, the subject of interpretation, and the world, a structure that determines what thoughts and utterances are actually about. The envatted brain “isn’t speaking the language” that the skeptic thinks it is (Marschall 2026, 285). We must interpret most of what the brain is actually saying, according to Davidson, as true.

My principal aim in this paper is to examine some of Davidson’s remarks about the scenarios discussed at the conference that seem difficult to reconcile with his standard presentation of triangulation. Chief among these is his curious suggestion that an envatted brain could be interpreted as perceiving and talking about the world outside its vat:

Right, if the scientists are observing the outside world and feed into the computer what’s going on in the outside world, then [the envatted brain is] in causal contact with the outside world. [It has] just got a certain more complicated perceptual apparatus than the rest of us. And then [it is] talking about the outside world. (Marschall 2026, 378)

Yet since the scientists in this case are acting as a mediate cause between the object in the outside world and the envatted brain when they “feed into the computer” the relevant data, the triangular figure Davidson elsewhere insists upon has been flattened into a straight line. Does the resulting picture still have enough structure to motivate the claim that the brain’s thoughts have particular content—and by extension, that the brain has beliefs at all? Answering questions like these will, I argue below, clarify the commitments of triangulation, and the conditions that Davidson thinks must be met by the occupants of its vertices.

In addition, working through the transcript sheds light upon what distinguishes Davidson’s antiskeptical argument from others in the family of semantic externalist responses to which it belongs. Epistemologists interested in this family have tended to focus upon the version Hilary Putnam presents in Reason, Truth, and History, which has been charged with missing the force of skeptical doubt, of providing a merely technical solution that, to borrow a Wittgensteinian turn of phrase, fails to give philosophy peace. The characters in Eli Hirsch’s philosophical dialogue about skepticism, for instance, object that Putnam doesn’t touch the anxiety felt by a person gripped by the possibility of skepticism, who they name Vatol. One character, Lev, is incredulous at the proposal that talking about the truth or falsity of the envatted brain’s utterances could be of any help to the skeptic:

Lev: Putnam’s argument is most terrible on beliefs about other people. If I am a vat-person then, for Putnam, I believe truly that “I discourse with other people who have feelings and thoughts about me.” Vatol, if he is smart, will not be anxious? How could this argument be honest and serious? (Hirsch 2018, 32)

On this point, I believe, Davidson’s antiskeptical argument fares somewhat better. After sketching his account of the role triangulation plays in the development of thought and speech in section one, in section two I formulate the antiskeptical argument he believes it supports. I turn to the conference transcript in section three, and after prying apart various skeptical scenarios that are suggested by the participants’ discussion, I contrast the ways that they would be treated by Putnam and Davidson. Ultimately, I shall argue, a distinct strength of Davidson’s approach is the solace it may provide someone gripped by a form of solipsistic anxiety, one who is primarily afraid of being alone.

1 Triangulation: Thought and Talk

Elements of triangulation can be detected in Davidson’s work on radical interpretation in the mid 1970s, but the importance of the triangular structure only comes to the forefront of his philosophy in the 1980s and 1990s. Commentators have been divided about how to understand the argument or arguments that Davidson bases upon triangulation.1 Yet most recognize that he appeals to more than one triangular figure. He introduces a primitive version of triangulation when discussing the coordination of social, nonlinguistic animals, and a second, more sophisticated version that obtains between animals capable of talking to each other. It is the more sophisticated form of triangulation that will primarily concern us, and the way that Davidson believes its structure to be necessary for the emergence of propositional thought.

Our thoughts, Davidson observes, have content, in the sense that they are about things. But what things? What determines what a given thought is about? Davidson describes events in the world as spawning “causal chains,” some of which impact our sensory receptors.2 Sunlight bouncing off a red apple may stream into the retina of my eye, for example. As I look at it, I might form the thought there is a red apple. But now, what makes my thought about the apple, rather than the image of the apple on my retina, or a point in space the light ray travels through between the apple and me, or a point in space the light ray traveled through before it hit the apple at all?

I cannot answer this question by insisting that I am thinking of the apple and not any of these alternatives, for that would simply beg the question: how is it that I am able to think of the apple, for my thought to have that as its content? Nor is this an empirical problem of establishing what sorts of things cause us to respond to them, which biology or psychology might help us to answer, but rather a conceptual problem of how we should understand the very idea of our beliefs having empirical content (Davidson 1992, 119).

One might suggest that a solitary individual could recognize perceptual similarities between their current and prior experiences, noting that their encounters with apples, say, share a certain perceptual quality. Could such a person “divide through” the additional qualities not shared by such experiences to locate the content of their current thought as about the apple? Not according to Davidson. He takes Wittgenstein (via Kripke) to have taught us that without an independent check, such an individual cannot properly be described as correct or incorrect in their judgment of similarity (Davidson 2001b, 3). The reason is that there is nothing in this picture to distinguish what currently seems similar to the individual from what is in fact similar for them. Davidson concludes that there is no principled way that a solitary individual could isolate a particular point in the causal stream as the content of their thought. And a thought without particular content is no longer a thought at all—at least, not a thought like those that we have.3

What distinguishes our own situation from that being imagined is that we are not solitary individuals. Instead, we are social animals that enter into triangular relationships with each other. Our “innate similarity responses” to our shared environment are themselves similar, meaning that we react to—and observe each other reacting to—the same sorts of distal events (Davidson 1992, 120). The causal chain that stems from the apple to my eye intersects with the causal chain stemming from the apple to your eye, precisely at the apple, and each of us distinguish the apple as a salient aspect of our experience. This triangular structural relationship, Davidson claims, is necessary for there to be an answer to the question of what our thoughts are about, and so, for our thoughts to have content.

This story about triangulation needs to be refined in various ways. First, a critic might object, if our thoughts about the world only have content when we are triangulating with another observer, how can we make sense of solitary observation? Surely a single witness to an event is able to form thoughts about it. Moreover, not all of our thoughts are about spatiotemporal objects. Causal chains do not stream out to us from theoretical constructs or abstract objects, and so there can be no triangular point of intersection.

These objections are based upon a misunderstanding of what triangulation is meant to establish. Davidson does not claim that triangulation is necessary for each of our thoughts to have the content they do. (Still less does he claim that triangulation suffices to give content to particular thoughts.) His claim rather is that triangulation is required for any of our thoughts to have content at all, and for us to have the status of thinkers. Triangulating with others “in the simplest cases” (Davidson 1997, 129), those concerning our mutual observation of objects in our shared environment, is meant to provide the structure against which propositional thought can emerge, if creatures successfully “make use of the triangular situation to form judgments about the world” (Davidson 1997, 130).

A second sort of objection is that Davidson’s account of triangulation fails to answer his motivating question. That question, recall, was how any point in the causal chain streaming from the world to a thinker could count as what the thought was about. But the causal chains that stream from events in the world to triangulators do not start at the point of triangular intersection. They stream further back into the causal past. Strictly speaking, then, the causal chains leading to triangulators intersect at multiple points. And so, the objection concludes, the thoughts of triangulators—if “thoughts” we may now wonder they have—still lack determinate content.

This is a deeper objection, requiring further reflection upon what Davidson packs into the idea of a causal chain, and it is one that the participants at the Stanford conference discuss in some detail. I will accordingly have more to say about it in section three, but as a first response we might emphasize a salient difference between the cases. There is no principled way to identify any point in the causal chain streaming to the solitary individual as the content of their thought. But there is a principled way to identify a point in the causal chain that streams to triangulators, namely, the initial point of intersection. Peter Pagin suggests that we might take this “closest common cause” to be the object of triangulators’ thought in typical cases (Pagin 2001, 202). Jason Bridges objects that Pagin’s suggestion is arbitrary, and that the closest common cause may in fact be further in than the object itself (such as air vibrations in our immediate vicinity that were caused by a distant tree falling) (Bridges 2006, 301). But is this arbitrary? Could social animals understand themselves to be responding to more mediate stimuli than the distal objects of which they are aware in the simplest cases of triangulation? We must return to these questions when considering the conference participants’ discussion of skeptical scenarios.

A third objection challenges the claim that the triangulation grounding thought must be linguistic. Recall that Davidson allows a primitive sort of triangulation that can obtain between social animals. Why does he deny them the more sophisticated version that allows for thought? Surely, a critic may object, two kittens playing together with a toy mouse are triangulating and thinking about the mouse?

Davidson allows that we may find it convenient to describe animal behavior in terms of propositional attitudes.4 In doing so, we are essentially triangulating with them, noting things in our shared environment to which we take them to be reacting. We say that our pets are afraid of the doorbell, not of the sound waves that sometimes emanate from the door. But the key question for Davidson is whether the pet is capable of understanding its situation in the way that we do. In general, when considering putative counterexamples to Davidson’s proposal, it is important to separate how things appear to us—creatures with the relevant capacities for thinking and talking about the world—from how things would appear to them. (This distinction will also matter when we come to imagine various envatted brains.)

Our capacity for thought has features that, according to Davidson, cannot be readily explained without linguistic triangulation. For in addition to being about the world—the feature upon which I have so far been focusing—we recognize that our beliefs may be false. As Davidson writes:

A creature may interact with the world in complex ways without entertaining any propositions. It may discriminate among colors, tastes, sounds, and shapes. It may learn, that is change its behaviour, in ways that preserve its life or increase its food intake. It may ‘generalize,’ in the sense of reacting to new stimuli as it has come to react to prior stimuli. Yet none of this, no matter how successful by my standards, show that the creature commands the contrast between what is believed and what is the case, as required by belief. (Davidson 1982, 104–05)

One way that Davidson picks out this feature of human cognition—our command of “the contrast between what is believed and what is the case”—is by attributing to us grasp of the “concept of objectivity,” for recognizing the contrast between belief and truth amounts to understanding that there is an objective truth beyond what one subjectively believes (Davidson 1997, 129). A being who does not possess the concept of objectivity, who lacks the ability to understand that it might be mistaken, lacks what we call thoughts. And linguistic triangulation illuminates how a being could acquire the concept of objectivity, Davidson thinks, because in communicating with another person and making sense of divergences between our separate experiences a “space” opens up in which “awareness of the possibility of error can take root” (Davidson 2001a, 293). If a person, that is, reckons with the mismatch between its own experience and that of another person by judging the other to have made a mistake about the world, then as they reflect on the situation they may come to realize that their own experience of the world could be mistaken. They thereby grasp the potential for a gap between how they believe the world to be, and the way the world truly is. Appreciating that neither it nor the being with which it triangulates has privileged access to the world they share puts the being into what we, looking on, recognize as the normative epistemic position. The triangulating being is now subject to having its views corrected by both its solitary recalcitrant experiences and the contrary reports of other people.5

Davidson’s celebrated argument from surprise—that our capacity for surprise involves a second order belief about our beliefs, and hence requires that we possess the concept of objectivity—predates his discussion of triangulation, but is, I think, strengthened if read in its context. Davidson describes the surprised person as aware that although she believed p, she now believes not-p, and thinks “such awareness … is a belief about a belief: if I am surprised, then among other things I come to believe that my original belief was false” (Davidson 1982, 104). Hans Glock objects to Davidson’s description of surprise, complaining that he “has no argument to rule out the possibility of a simply switching from a belief that p to a belief that q, without that switch involving a believing that its original belief was false” (Glock 2003, 287, original emphasis). I might add that sometimes we are surprised because our new belief was unlikely on the basis of our current beliefs, not because we previously believed it false. But Davidson need not hold that everything we call surprise involves a belief about a belief.6 It suffices to point out that if a being were incapable of experiencing the sort of surprise he has identified, we should not call its attitudes “beliefs” at all. We would have no way to understand a as recognizing the difference between the world as it currently appears and the world as it in fact is, and it is only in the recognition of this difference—which, if he is correct, involves linguistic triangulation—that our concept of “belief” finds purchase.

The final two objections I want to consider concern the upshot of Davidson’s reflections upon triangulation. First, one might concede that possessing the concept of objectivity is necessary for any creature capable of what we call thinking, and that linguistic triangulation is an explanation for how it could be acquired. But why should we accept that this is the only explanation? Davidson’s report that he has “[no] idea” how “one could come to have the belief-truth contrast” other than through communication with another person leaves this possibility open (Davidson 1982, 104). Second, many commentators have objected to the circularity of Davidson’s account. For how can triangulating creatures acquire the concept of objectivity by coming to judge that others are making mistakes about the world, if “coming to judge” is itself a species of thought that requires the concept of objectivity? Davidson acknowledges his view that “unless the creatures can communicate, unless they can engage in the exchange of propositional contents, there is no way they can take cognitive advantage of their ability to triangulate their shared world” is an “unhelpful answer” to the question of what enables creatures to move from the primitive to the complex triangulation situation (Davidson 2001a, 13).

To some, these last two objections rob triangulation of its interest. Jason Bridges, for instance, concludes that Davidson’s suggestion “that the transition into thought is effected by a creature’s cottoning on to the fact that it and another creature are jointly triangulating on particular events in the world” cannot “in any sense be art of the explanation of the acquisition of the capacity for thought, given that it presupposes that capacity” (Bridges 2006, 301).7 But others have found the circularity virtuous and the onus regarding how beings might otherwise acquire the concept of objectivity to rest with Davidson’s critics. Claudine Verheggen, for example, having emphasized that Davidson’s goal is to show how triangulation makes thought possible rather than to explain in detail how beings acquire the ability to think, concludes that his is “obviously not just the narrow vicious circle that says there is language and thought only when there is language and thought,” but rather “the rich and complex circle that encompasses language, thought (each of which depend on there being fixed meanings), possession of the concept of objectivity and linguistic triangulation” (Myers and Verheggen 2016, 35). Triangulation does not offer, and nor did Davidson think it offered, a reductive account of intentional thought in non-intentional terms (Davidson 1997, 139). Those hoping for such an account may well be frustrated by triangulation, but Verheggen maintains that tracing the circle he draws may nevertheless be “instructive.”8

Although I cannot pursue this in detail here, and despite Davidson’s protestations about the “dearth of arguments” to be found in the history of philosophy about why thought requires language (Davidson 1982, 101), I think that considering the antecedents of triangulation in the American pragmatist tradition only adds to the instructiveness of the circle that impresses Verheggen. George Herbert Mead’s account of the emergence of objective thought among social creatures dovetails with Davidson’s. For Mead, it is “the social process, through the communication which it makes possible among the individuals implicated in it” that “is responsible for the appearance of a whole set of new objects in nature, which exist in relation to it (objects, namely, of ‘common sense’)” (Mead 1934, 79, my emphasis). For Mead, like Davidson, the reason epistemology must begin in a social setting is that it is only in that setting that the objects of our thought exist. Mead also draws a firm line between the cognition of humans and other animals, on the grounds that animals are not conscious of the “objects” that they experience (Mead 1934, 65).9 The distinctive “role-taking” cognitive mechanism Mead attributes to humans, which allows them to project themselves into another’s position and ultimately develop objective thought about a shared world, is the ancestor of Davidson’s triangulators appreciating the significance of their diverging experiences (Mead 1934, 184–85).10

Let us now turn from the insights about thought and talk Davidson credited to triangulation, to the relevance of triangulation to the problem of skepticism.

2 Triangulation and Skepticism

Davidson did not set out to repudiate skepticism (Davidson 1999c, 163; 2001a, 289). Rather, he came to appreciate that his approach to the philosophy of mind and language had antiskeptical consequences.11 In early papers, Davidson had argued that it was in the nature of belief to be veridical, since in interpreting a person as having beliefs at all one was obliged to take most of their beliefs to be true (Davidson 1983, 149–50). Since this position rules out the possibility of a person being massively wrong in their beliefs, one cannot be a global skeptic who doubts that their own beliefs are true. His late work on triangulation allowed him to present a different version of this argument. For our beliefs to have the content that they do, he now argued, we must have triangulated with another person in a shared environment. We cannot, therefore, intelligibly doubt that that environment, or the other people in it, are largely as we believe them to be.

Before considering criticisms of Davidson’s approach, it will be helpful to formulate his reasoning more carefully:

P1. I have objective beliefs about the world that are either true or false.

P2. In order to have objective beliefs about the world that are either true or false, my beliefs must have particular content and I must possess the concept of objectivity.

P3. In order for my beliefs to have particular content and to possess the concept of objectivity, I must have entered into causal, linguistic triangulation with another person.

P4. If I have entered into causal, linguistic triangulation with another person, the external world and other people exist.

C. The external world and other people exist.12

Davidson takes P1 to be the Cartesian starting point, for a skeptic worried that their beliefs about the world might be false must at least believe that their beliefs are either true or false, and also that their beliefs are about the world. Davidson’s arguments that beings with thoughts like ours must have the concept of objectivity and that thoughts like ours have particular content supports P2. His account of linguistic triangulation as the way we acquire the concept of objectivity and have thoughts with particular content grounds P3. P4 elaborates an apparent consequence of linguistic triangulation, and the antiskeptical conclusion C follows.

Given the Cartesian starting point of this reconstruction, it is worth contrasting it with Descartes’ cogito. Instead of “I think, therefore I am,” Thomas Nagel describes Davidson as arguing “I think, therefore I know” (Nagel 1999, 196). Applauding the wittiness of Nagel’s description (since its French articulation “je pense, donc je sais” differs from Descartes’s “je pense, donc je suis” in just a single letter), but wanting to emphasize that his conception of knowledge is based in triangulation, Davidson proposes instead “I think, therefore you are” (Davidson 1999b, 207). I agree with Nathaniel Goldberg’s suggestion that “I think, therefore you are too” is better still (Goldberg 2022, 8), but ultimately recommend “I think, therefore we are.” For Davidson, our epistemology has gone wrong if it ascribes any priority to the first person position (as is suggested by the structure of “I___, therefore you___”). If we internalize the lessons of triangulation and recognize the conditions that must obtain for our thoughts to have the nature they do, he thinks, we must build our theory of knowledge from an avowedly social position. As he put it in an instructive 1990 interview,

The empiricists have it exactly backwards, because they think that first one knows what’s in his own mind, then, with luck, he finds out what is in the outside world, and, with even more luck, he finds out what is in somebody else’s mind. I think differently. First we find out what is in somebody else’s mind, and by then we have got all the rest. Of course, I really think that it all comes at the same time. (quoted in Borradori 1994, 50)

How successful is Davidson’s antiskeptical argument? An obvious weakness is P3. As I noted in the previous section, Davidson’s claim that causal, linguistic triangulation gives an explanation for how individuals could acquire the capacity for propositional thought does not support the claim that it is the only such explanation. And some have thought that the most Davidson’s reflections on interpretation could establish is that a person must believe that they have entered into causal, linguistic triangulation with another person, not that they have done so (see, e.g., Genova 1999, 187; and Lepore and Ludwig 2005, 403). Aren’t these further guarantees required for the argument to have any force against the skeptic?

Davidson concedes that he has “no knock-down argument against skepticism” (Davidson 2001a, 288). But he still believes that he has responded to the skeptic. For if one adopts his account of the conditions under which propositional thought emerges, one comes to see that one’s view of the world must be largely correct, thereby occupying a position where “there is no point in attempting, in addition, to [showing] the sceptic wrong” (Davidson 1999a, 163, original emphasis). Joining Richard Rorty, elsewhere he bluntly tells the skeptic to “get lost” (Davidson 1987, 157).

Sympathetic commentators have sought a less combative way to present Davidson’s proposal. Bjorn Ramberg describes him as “issuing an invitation” to adopt the “radical” position that thought and agency depend upon linguistic triangulation, in part because of its antiskeptical power (Ramberg 2001, 232). Duncan Pritchard suggests that Davidson is providing a “therapeutic” response to skepticism that exposes why it can be treated as a “pseudo-problem” (Pritchard 2013, 529). A different strategy is to marshal Davidsonian insights into a new antiskeptical argument. Barry Stroud argues that Davidson has shown “belief-attribution is in its nature (largely) truth ascribing,” which suffices to ground a modest transcendental argument with antiskeptical force (Stroud 1999, 149). In Stroud’s view, skeptical alternatives are “not [possibilities that we can] consistently believe to be actual” (Stroud 1999, 157), for Davidson has shown that in attributing beliefs to ourselves we must believe those beliefs to be true. We cannot simultaneously believe that skeptical scenarios—which Stroud concedes are logically possible—obtain.

Yet other commentators have not been persuaded. Even if Stroud’s modest transcendental argument could convince us that we cannot consistently believe skeptical scenarios are true, might we not still be alarmed by the bare logical possibility that one of them is true?13 And, recalling Hirsch’s dialogue, how inviting will Vatol the anxious skeptic find Davidson’s “invitation,” how effective his “therapy”? As John McDowell puts it:

The Davidsonian response seems to be that if one were a brain in a vat, it would be correct to interpret one’s beliefs as being largely true beliefs about the brain’s electronic environment … [this] response does not calm the fear that our picture leaves our thinking possibly out of touch with the world outside us. It gives us a dizzying sense that our grip on what it is that we believe is not as firm as we thought. (McDowell 1994, 16–17).

These sort of doubts, I think, actually take aim at P4. Supposing I have entered into causal, linguistic triangulation with another person, that is, what kind of guarantee do I have that the external world and the other person exist in anything like the way I thought them to? Might we both be envatted brains triangulating in an artificial environment—and isn’t that something to be alarmed by? “Skepticism seems to have been repressed and rendered mute, rather than really refuted,” complains Nagel, “it is still there, trying to find an outlet” (Nagel 1999, 203). In order to get clearer about this question and how Davidson would treat it, let us turn to the envatted brains discussed at the Stanford conference.

3 A Series of Envatted Brains

Readers who are familiar with the transcript might be raising an eyebrow at the title of this section, for the conference participants are dubious about the philosophical value of thought experiments laced with science fiction. Dreben insists that their conversation should be grounded in what he calls “serious science,” and thinks that they should refuse to be goaded into giving substantive answers to questions that only call for “semantic legislation” (Marschall 2026, 175–76, see 44). Dreben’s idea is that some of the words used in describing a theory may lack clear application criteria in sufficiently outlandish cases—when language has gone “on holiday,” as Wittgenstein might say (Wittgenstein 1953, sec. 38)—so that although in such contexts a decision might perhaps be made to fix their meaning in a given way, they do not currently have a meaning to which our theory is beholden. In the idealized case of a skeptical scenario said to accommodate “all possible observations,” Dreben complains, “observation” has become detached from its ordinary scientific use (Marschall 2026, 175). Scientific observations are always incomplete and actual. Accordingly, the scenario does not ground an objection that necessitates a response from a philosopher, like Quine, whose own theory appeals to observation (in the form of observation sentences).

Quine is taken by this maneuver for sidelining cases where “a real science fiction question is raised in philosophy” (Marschall 2026, 177, see 187). Føllesdal likewise says of radically skeptical proposals, such as Bertrand Russell’s suggestion that “we could assume that the world was created five seconds ago with all the fossils and all the memory traces and so on,” that “we don’t want to take that too seriously” (Marschall 2026, 187). As a result of their united front, wherever possible, Davidson—who in the context of the conference is primarily motivated by his desire for the others to engage his own objections to Quine’s epistemology—tries to distance his position from science fiction.14 “Now we come to my main point,” he remarks, “it doesn’t necessarily have much to do with science fiction though sometimes that helps to make clear what the point is” (Marschall 2026, 188).15

Yet, despite the participants’ qualms, since my own aim is to clarify Davidson’s position rather than his critique of Quine, I think it is worth probing further into the science fiction of envatted brains and asking why Davidson believes that they may help him clarify his point. In doing so, it is crucial to describe the brains with some care. As Davidson puts it, “Well I think a lot does depend on the details when we try to do these funny things” (Marschall 2026, 375, see 254). Upon what basis, for instance, do the scientists manipulating the envatted brain judge that it is thinking (Marschall 2026, 224)? How does the brain get feedback from the scientists about its simulated environment (Marschall 2026, 226–27)? How does the brain learn to communicate (Marschall 2026, 249)? Much of what Davidson thinks he can say to the skeptic will turn upon the answers we give these questions. So let us consider the first in our series of envatted brain cases.

KIDNAP. While taking a stroll on a beach in Miami, Frank is knocked unconscious, kidnapped, and taken to the laboratory of Elena, a diabolical—and diabolically advanced—neurophysiologist. Elena surgically removes Frank’s brain from his body and places it in a vat of nutritious fluids. She then proceeds to connect Frank’s brain to a supercomputer, HAL, which is capable of systematically stimulating the neurons in his brain in just the way they had previously been stimulated by the nervous system of his body. She programs a simulation of the Miami beach from which Frank was kidnapped into HAL, which HAL converts into electronic signals that it sends to Frank’s brain. Frank’s brain begins to systematically respond, and on the basis of this feedback HAL sends further, updated electronic signals. Elena watches a monitor attached to HAL that displays a representation of the audiovisual component of the signals it is currently sending to Frank’s brain. As she watches the sights and sounds of the Miami beach pass by on the screen, she judges that Frank has experienced no discontinuity: unaware that he has been kidnapped, he still believes himself to be enjoying his stroll.16

KIDNAP may be thought to ground a skeptical challenge. If Elena’s interpretation of Frank’s experience is correct, then although he believes himself to be strolling the Miami coastline, he is actually trapped in Elena’s laboratory. And Elena could have chosen a different target. Might we be trapped in her laboratory?

At the conference, Davidson makes clear that he has no response to this type of skeptical challenge. “I divide brain in the vat problems into two completely different ones,” he says, “one is where the brain is stolen in the middle of the night and has been brought up as we all have, and in that case I just think it’s going to be massively deceived” (Marschall 2026, 253, see 223–24, 226). Putnam, too, would have nothing to say about KIDNAP (Putnam 1981, 16 n3). For both Davidson and Putnam, newly envatted brains that were previously in causal contact with the external world are fully capable of thinking and talking about that world, and so, can be correctly described as mistaken about their true, disembodied situation.17

How alarmed should we be by this failure to answer KIDNAP? Absent news stories of Miami kidnappings, depraved advanced scientists, or supercomputers, it would seem premature to take this very seriously. Were Hirsch’s character Vatol made genuinely anxious by reflecting on this science fiction, we should probably refer him to a therapist or give him a romance novel. But the story does serve as an important reminder that Davidson did not believe or represent himself as answering all skeptical scenarios. Rather, he thought triangulation undermined a particular class of scenarios, and that seeing how and why it did so could be philosophically fruitful. So let us consider a second envatted brain case that is squarely within his target.

SUPERCOMPUTER. Reality consists of billions of disembodied brains floating in a vat of nutritious fluid, attached to an enormous supercomputer, HAL+. Periodically, new brains spontaneously grow in this fluid, while others shrivel and die. The enormously complex program running on HAL+—a program that was not programmed by a diabolical scientist, but that has been running for all time—sends electronic signals to and receives feedback from each brain in the vat. The signals stimulate the brains’ neurons in just the way that they would be stimulated if they were embodied, taking with each other and moving about in a shared world. HAL+ perfectly coordinates the brains’ responses in its simulation, so that if one brain experiences itself as “speaking” to a second, the second brain experiences itself as “hearing” the first. Although it seems to the brains that they are embodied, talking with each other and moving about in a shared world, they are all massively deceived.18

Like KIDNAP, SUPERCOMPUTER may be thought to ground a skeptical challenge. If the story is consistent, what reason do we have to believe that we are not disembodied brains being manipulated by HAL+?

According to Davidson, SUPERCOMPUTER is not consistent. The envatted brains cannot be described as being “massively deceived,” for the content of their beliefs is what is causing them to be stimulated. When the brains triangulate with each other and acquire the concept of objectivity, their causal chains intersect within HAL+’s simulated environment. If speakers attached to HAL+ were to give the audio output of what one of the envatted brains was “saying,” it might sound identical to a sentence of our language, like “there is an apple” (see Marschall 2026, 254). But homophony is not synonymy. Our “apple” is about apples, but the envatted brain’s “apple” is about HAL+’s simulations of apples. Accordingly, the thoughts and utterances of the brains in SUPERCOMPUTER are largely true. Putnam would be slightly more cautious, allowing that different theories of meaning might deny that the brains “utterances” mean anything, or else locate the brain’s “apple” as referring to the electronic impulses HAL+ is sending the brains or some feature of HAL+’s program (Putnam 1981, 14). Yet he would ultimately concur with Davidson’s conclusion that the brains are not massively deceived.

SUPERCOMPUTER also grounds the sort of skeptical challenge that Davidson urges the conference participants to see that Quine cannot answer. Whereas Davidson takes shared distal causes in the world to fix the meaning of triangulators’ thought and speech, Quine’s proximal view focuses upon the way each individual is stimulated by the world. “This proposal of Don’s has advantages when it comes to answering skeptics and so on,” Føllesdal remarks, “But you [Quine] still don’t feel quite happy with shifting to that though.” “No,” Quine replies, “I guess for me the meaning so to speak hinges on the irritations of the surfaces, and for Don it hinges on what’s out there in the world” (Marschall 2026, 272). Since the brains in SUPERCOMPUTER are stimulated in just the way that our brains are stimulated in reality, Quine interprets them as speaking the same language as we do. But he remains unfazed by the skeptical challenge this might seem to ground, based as it is in science fiction (Marschall 2026, 224).

As a result, Davidson turns to more humdrum cases to press his attack. Quine cannot be right that sensory stimulations are the basis of meaning, he thinks, because if they were we would end up mistranslating a person who was stimulated differently by the world. “The point that I keep coming back to” Davidson insists, is that “on [Quine’s] method of translation a person can be talking about all sorts of things with which he has never had any contact whatsoever” (Marschall 2026, 370). In what is recognizably an early version of the warthog argument he would later publish (Davidson 1990), Davidson imagines a person whose brain is stimulated in an unorthodox way:

Whenever there’s a rabbit there he had a certain response, whenever there’s a book there he has a certain response. But the patterns of stimulation are such that the patterns of stimulation of books for him is the same as the patterns of stimulations of rabbits for me. (Marschall 2026, 371–72)

If we knew that a person’s brain were wired in this unusual way, Davidson thinks it obvious that when we encounter a rabbit together we should still take his utterance “there is a rabbit” to mean that there is a rabbit. The rabbit is the shared distal cause of both his and our own utterance. But Quine must instead say that the person’s utterance “there is a rabbit” actually means that there is a book (since that is how we know he is being stimulated), and accordingly reckon him as wrong about the world. To Davidson, this result flies in the face of “not only … common sense, but even to what any good translator would do” (Marschall 2026, 370).

In the years after the conference, Quine became persuaded by Davidson’s critique and abandoned the idea that the world could cause people with unique neurophysiologies to have the “same” pattern of stimulation. “What floats in the open air is our common language,” he later writes, “which each of us is free to internalize in his peculiar neural way” (Quine 1992, 44). Private stimulations do not matter for semantics, he came to believe, as our capacity for language is explained by our having evolved to find similar aspects of the shared world salient, and by our capacity for empathically projecting ourselves into another’s point of view. Nevertheless, he insists that stimulations remain crucial for the epistemological project of accounting for how an individual acquires knowledge of the world.19

During the conference, however, the participants remain skeptical of Davidson’s proposal. Dreben especially is concerned to keep Quine’s philosophical project separate from Davidson’s, allowing that although both may be interesting, they have fundamentally different aims and constraints (Marschall 2026, 361–62). According to Dreben, Quine has clearly defined his “basic technical notion [of] stimulus meaning,” used it to give a credible, robust account of how certain sentences within a linguistic community could count as “observation sentences,” and proceeded to explore the consequences of the meaningfulness of language that results (such as the indeterminacy of translation) (Marschall 2026, 234–35). In contrast, he finds the basic notions that Davidson is relying upon—events and causal chains—perilously vague. “You are not interested in giving, as far as I can tell, clean explications of almost any of the notions that you used,” Dreben challenges Davidson (Marschall 2026, 232). Føllesdal is similarly doubtful about Davidson’s reliance upon “causal chains,” suggesting that “causal trees” would be more accurate, noting “the problem again is [determining] what events in that big tree would we be talking about” (Marschall 2026, 255; see Føllesdal 1999). The tenor of these and other remarks suggest that both Dreben and Føllesdal thought Davidson’s philosophical project lacked necessary detail, at least in comparison to Quine’s.

Before turning to a third envatted brain, let us consider what Hirsch’s character Vatol would make of the Davidsonian response to SUPERCOMPUTER. The inconsistency in the scenario is in describing the envatted brains as massively deceived, but might Vatol nevertheless be anxious that, even though the brains have true beliefs about their world, they are not aware that they are envatted? Similarly, he may frantically point out, even though his beliefs about the world are true, he could be unaware about—and forever unable to articulate the nature of—real reality at a deeper level of description. As Graeme Forbes writes, “If ideally rational inquirers can be Putnam-style brains in a vat, then even though their beliefs are not generally false, and so may aspire to the status of knowledge, there are truths that they will never know, which means that truth outstrips the reach of ideally rational inquiry” (Forbes 1999, 72).

Yet seriously worrying about the nature of real reality beyond what we can ever think or say, I would try to counsel Vatol, is to refuse the finitude of the human condition. Any constraints we might discover on ideal rational inquiry should not undermine the value of our own more modest investigations. Better to focus on the positive result of reflecting on SUPERCOMPUTER—the fact that we are communicating with others about the world we do share, and that our beliefs are largely true.

Davidson and Putnam would give the same answers to KIDNAP and SUPERCOMPUTER. But their views would come apart, I think, regarding a third case.

SOLIPSISM. Reality consists of a single disembodied brain floating in a vat of nutritious fluid, attached to an enormous computer, HAL. The program running on HAL—a program that was not programmed by a diabolical scientist, but that has been running for all time—sends electronic signals to and receives feedback from the brain in the vat. The signals stimulate the brain’s neurons in just the way that it would be stimulated if it were embodied, talking with other people and moving about in a shared world. Although it seems to the brain that it is embodied, talking with other people and moving about in a shared world, it is massively deceived.20

Putnam would treat SUPERCOMPUTER and SOLIPSISM similarly. As I noted above, he cautiously allows that the meaning of what envatted brains “say” will differ depending on different theories of meaning, and that on some they may not mean anything at all (Putnam 1981, 15). What he is positive that they do not mean, though, is what we, unenvatted people mean by our words, since we are in contact with the world and the envatted brains are not. If the brains in SUPERCOMPUTER or the brain in SOLIPSISM have beliefs, Putnam would conclude, those beliefs are about something in the computer simulation.

But armed with his theory of triangulation, Davidson wants to deny that SOLIPSISM is intelligible. The brains in SUPERCOMPUTER acquire the concept of objectivity by triangulating with each other in HAL+’s simulation. But the brain in SOLIPSISM cannot do so. “I would say this is an unthinkable possibility,” he says during the conference, “If you find there’s nobody else in the world, then you find that after all you don’t mean anything. You don’t think anything” (Marschall 2026, 358).

This is a surprising conclusion that is difficult to sustain. HAL+ might stimulate a brain in SUPERCOMPUTER in exactly the same way that HAL stimulates the sole brain in SOLIPSISM, so how could Davidson be entitled to judge the cases differently? In particular, what prevents the sole brain in SOLIPSISM from triangulating with simulations of other people in HAL’s program, acquiring the concept of objectivity as a result of believing it has talked to other people? Bridges objects that the “social aspect of Davidson’s story is functionless,” (Bridges 2006, 295) since what ultimately does the work in triangulation is not the shared experience of triangulating beings, but each triangulator’s ability to grasp the significance of its own experience. Likewise, Ernie Lepore and Kirk Ludwig believe that triangulation only warrants “the conclusion that we must think of ourselves in communication with others in order to have the concept of belief, truth, and objectivity (Lepore and Ludwig 2005, 403), not that we are in communication with others.

The response that I think Davidson would like to give here is to insist that the envatted brain in SOLIPSISM lacks the concept of objectivity, and so cannot be credited with having what we call beliefs. We should not give into the temptation to say that, since its simulated experience is like that of talking to others it can acquire the belief/truth contrast, because that contrast is only applicable in genuinely social situations. Our thoughts only come to have particular contents via triangulation, and so its solitary situation prevents it from having what we call thoughts. Might Vatol anxiously complain that he does not know if what seems to him to have been social experiences are genuine? Perhaps, like the brain in SOLIPSISM, he only seems to have had triangular interactions with others, and so doesn’t actually have beliefs. But to Davidson, this anxiety is unsustainable because it denies what even the Cartesian skeptic must accept. P1 of the argument I reconstructed in the previous section is that I have objective beliefs about the world that are either true or false. The skeptic relies upon this premise being true so that he can doubt his beliefs, wondering if they are actually false. We can counsel Vatol that he cannot merely be believing himself to have beliefs, because doing so, after all, is already to have a belief.

I suspect that neither Bridges nor Lepore and Ludwig would find this response persuasive. It is hard to see why the brain described in SOLIPSISM could not win itself a concept sufficiently like the genuine concept of objectivity with which to form something much like our beliefs, which may engender a corresponding skeptical anxiety. But there is another strategy that I think a Davidsonian might gainfully pursue. Suppose we grant that the brain in SOLIPSISM acquires the genuine concept of objectivity; might Vatol nevertheless take some comfort from Davidson’s philosophy?

To see why, consider how the brain in SOLIPSISM learns its language. According to Davidson’s theory of triangulation, there must be simulations of other people In HAL’s simulated environment to which the brain’s simulated body can respond. The brain must also be able to observe the simulations of other bodies reacting to the simulated environment that they share. In virtue of attributing points of view to the simulated other people and appreciating that they may be mistaken about the shared world, the brain is able to acquire the concept of objectivity. The key thing to appreciate is the amount of feedback that is required for this story to make sense. HAL’s simulations of other people must be dynamic, systematically altering their behavior in response to what the brain is doing.21 So there is a real sense in which other people do exist in this version of SOLIPSISM; it is just that they are virtual rather than physical.

Part of the reason that the characters in Hirsch’s dialogue find Putnam’s response to skepticism unsatisfying is that they take it to be consistent with a lack of meaningful relationships with other people. An envatted brain may think it has fallen in love, has a family, shares intimate thoughts, and so on—but actually be mistaken. Putnam’s response leaves the characters with a distinctive sort of anxiety about the contingency of their own lives with those they care about. They may be wholly alone. Yet I think a Davidsonian might persuade Vatol to resist this line of reasoning. For the brain in SOLIPSISM does—indeed, must—have meaningful encounters with the virtual people in its simulation for it to acquire the concept of objectivity. Were it made aware by HAL that it was the only person with a physical brain, it would doubtless be surprised. But it should not conclude that its interactions and life with the virtual people around it might as well never have happened. The community in which it has lived is an irreplaceable part of what has allowed it to become the epistemic individual that it is.

The envatted brains in KIDNAP, SUPERCOMPUTER, and SOLIPSISM are all supposed to be ignorant of their disembodied situation. But the final two cases suggested by the conference participants’ discussion are somewhat different, as they concern interaction between those outside and inside the vat.

ROBOT. Due to an unprecedented birth defect, doctors discover that only the brain of Betty’s fetus is viable. However, medicine has advanced to such a degree that the brain can be safely removed from the cranial cavity, and placed in a vat of nutritious fluids that is attached to a supercomputer HAL. Robotics has also advanced: HAL is capable of wirelessly transmitting signals to and receiving signals from a robotic humanoid body, DATA, that is equipped with cameras instead of eyes, microphones instead of ears, tensile sensors all over its robotic skin, and biological sensors instead of an olfactory system. A beneficent scientist programs HAL to systematically stimulate the envatted brain in just the way that DATA’s sensoria is being stimulated at any particular moment. As the brain begins to respond to HAL’s stimulations, HAL codifies and transmits the feedback to DATA, causing the robotic body to move about. Betty and her family treat DATA as a baby, and eventually the speaker in its mouth produces audible speech. In the years to come, Betty and the brain are able to have conversations.22

This is the sort of case Davidson is imagining when he raises the possibility of indirect causal chains, and our ability to triangulate with an envatted brain. “If the story is just that there’s a complicated connection, but it’s a systematic one, then I’m going to say he’s talking about the things that his talk is systematically connected with,” he says, “And that will be those things in the outside world. As just as if one of us lost his eyesight but was hitched to a television machine that printed the stuff out on his chest” (Marschall 2026, 378). Applying triangulation, we find that the causes common to Betty and the brain—what their conversations are about—are objects and events in the real world. The causal chains streaming from the world hit Betty directly through her sensory organs, and hit the brain indirectly through DATA’s sensoria and HAL’s stimulations. This envatted brain has acquired the concept of objectivity and learned the language spoken by Betty and her family.

Far from grounding skeptical doubt, ROBOT belongs to a different genre of optimistic science fiction, exploring the promise of medicine and technology to improve people’s lives.23 We could even imagine Betty taking a teenage DATA into the laboratory to see HAL and the envatted brain without raising the specter of existential dread, as opposed to curiosity. Yet if we accept that this science fiction is intelligible, there is a closely related story we might find troubling. Føllesdal presses this nagging concern against Davidson repeatedly at the conference. “It’s not so clear that these surrounding people would actually be with that brain the way we’d be with our children” (Marschall 2026, 253), he worries: “you [Davidson] seem to have to take for granted … that the people surrounding the brain are really very truthful to him” (Marschall 2026, 250). Let us consider our final envatted brain case.

DISTORTION. Elena, a diabolical—and diabolically advanced—neurophysiologist, has removed the brain from the cranial cavity of Betty’s fetus and placed it in a vat of nutritious fluids that is attached to a supercomputer HAL. HAL is capable of wirelessly transmitting signals to and receiving signals from a robotic humanoid body, LORE, that is equipped with cameras instead of eyes, microphones instead of ears, tensile sensors all over its robotic skin, and biological sensors instead of an olfactory system. Elena programs HAL to systematically stimulate the envatted brain in a radical distortion of the way that LORE’s sensoria is being stimulated at any particular moment. If LORE’s cameras alight on an apple, for instance, HAL stimulates the brain as if it were looking at a banana. As the brain begins to respond to HAL’s stimulations, HAL codifies and transmits the feedback to LORE, causing the robotic body to move about the laboratory. Elena treats LORE like a child, and eventually the speaker in its mouth produces audible speech. In the years to come, Elena and the brain are able to have conversations, although Elena ensures that the brain is radically deceived about the true nature of apples, bananas, and much more of reality.24

This science fiction story illustrates why embracing the possibility of triangulation via indirect causal chains fuels further skeptical doubt. Having judged Betty in ROBOT to successfully talk with the envatted brain about the world outside the vat, we seem forced to accept that Elena in DISTORTION can do so too. For when we apply triangulation, we find that the causes common to Elena and the brain—what their conversations are about—are objects and events in the real world. The causal chains streaming from an apple in her laboratory, say, hit Elena directly through her sensory organs, and hit the brain indirectly through LORE’s sensoria and HAL’s distorted stimulations. This envatted brain therefore means what Elena does by its use of “apple,” But would deny such truths as “apples are roughly spherical.” The threat of skepticism has returned.

At the conference, Davidson tries to suppress this objection by appealing to the broader constraints the participants agree are already placed upon interpretation. “No, I’m not assuming [that people surrounding the brain are being truthful]” he interrupts Føllesdal, “As in Van’s and in my view, you’ve got to be able to figure out when people really have a certain attitude for a sentence. If you can’t do that you can’t get anywhere” (Marschall 2026, 250). But while it may be true that the project of interpretation presupposes the ability of detecting attitudes that others have to the world, it is far from clear that this blocks the threat posed by DISTORTION. So long as the envatted brain in Elena’s laboratory can determine the attitude she seems to have to their shared environment, nothing prevents it from learning its language by triangulating with her. For instance, Elena may point to an apple and say, “That is an apple.” Causal chains from the apple hit LORE’s sensoria, are manipulated by HAL, and appear to the envatted brain as a banana. The envatted brain agrees. Elena may now continue, and—lying—say, “apples are not roughly spherical.” The envatted brain, examining the non-spherical banana it perceives, may enthusiastically agree. In this way, it comes to be deceived about its environment.25

The irony here is that this skeptical challenge is the inverse of the one that we saw Davidson wanted to press against Quine. Davidson took it to be obvious that stimulations should not matter to translation: the person we knew to be book-stimulated upon seeing rabbits should be interpreted as talking about rabbits, not books, when they say things like, “that is a rabbit” as a rabbit hops by. But in a case like DISTORTION, he wants to pull us in the opposite direction. Despite the causal chain that runs from apples in the world to this envatted brain, he would want to interpret the brain as not talking about the external world:

If I want to interpret [an envatted brain] I’d have to do it in the usual way. I’d have to see what is causing it to assent to and dissent from the things that it says. Never mind that it sounds like English. That’s irrelevant. Now, either I find something or I don’t. If it’s hitched up to a computer and I can’t see what’s inside the computer, I can’t do anything. I’m not going to make sense out of this at all. But suppose that the computer is a good old fashioned one full of pulleys and wheels and so forth, and I can see these pulleys and wheels working. And so I notice that when this pulley goes it assents to so and so, and when it doesn’t go it doesn’t assent. So my first best guess is that when it says “there’s a rabbit,” what it means, what I should interpret it as meaning, how I should translate that is, the pulley is going. (Marschall 2026, 254)

Applied to DISTORTION, the idea would be to redescribe how Elena interprets the envatted brain. HAL’s stimulations are what is causing the brain to have beliefs, and the feedback HAL is sending back to Elena via LORE are what is causing her to judge herself to be communicating with the brain. So the triangle’s third vertex must be within HAL’s simulation. Although Elena is capable of using the word “apple” with its ordinary English meaning, having learned its application with other people outside the laboratory, in her conversation with the envatted brain she should interpret its use of “apple” as being about something in the machine—and by extension, she should interpret its beliefs as largely true.

The difficulty, however, is how to justify this proposal—and in particular, how to rule out DISTORTION without ruling out ROBOT. And the notion of “causal chain” to which Davidson frequently appeals seems neither clear nor robust enough to do this work. At the conference, Føllesdal is especially persistent in questioning Davidson’s easy invocation of “the” common cause as constituting the content of our speech. “The object that we are talking about could not be closer than the branching point,” he allows, “It has to be a common cause. But of course there could be common causes beyond the branching point” (Marschall 2026, 266). A little later, he describes a clear example: “Of course if you go to an art gallery you’re talking about the picture itself. But if you look at a photograph of a person you’re normally talking about the person rather than a photograph.” (Marschall 2026, 267). Davidson accepts that this grounds a problem, but does not seem to recognize its depth:

Davidson: Yes. That creates a problem; you’re right. The point where they first merge is certainly not where they last merge, where they’re last together. So it’s not the point at which the mergence begins.

Quine: Usually it is I would say.

Da: I agree. Absolutely.

Q: But certainly it wouldn’t be where they first converge. That gets us back to particles again … it’s this trouble again of limiting it to objects, books, rabbits, etc. rather than more generally situations or events, raining. (Marschall 2026, 267–68)

It is not obvious how to reconcile Quine’s remarks here. But what I think that Quine and Davidson are in agreement about is that the capacity to recognize representations (such as Føllesdal’s photographs of people) must come after the capacity to recognize presentations. The initial stages of learning a language, both agree, require empathetic projection on the part of both the parent and the child, and ostensive definition about things in their shared immediate environment.26 Yet Quine, like Føllesdal and Dreben—and Bridges, as we saw in Section 1—is suspicious of the empirical claim that Davidson seems committed to. It may well be common sense to interpret other people as talking about medium-sized dry objects in our vicinity rather than more esoteric alternatives, but “causal chains” do not seem capable of individuating the events and situations we encounter in the world. As Dreben drily summarizes Davidson’s proposal a little later,

Prima facie it seems quite plausible. But that’s because that’s really Moorean common sense to start with. You start saying that you’re going to start talking about a series of books in front of me, and, you see those books in front of me and I say, “there are books in front of me.” You begin to interpret me. That’s fine. But it doesn’t tell me very much. So at least you have to say quite a bit about the causal relation that’s going on here. And it just can’t be taken as primitive just like that. And then I ask, are you going to give me some physicalist account of his relation between the books in front of me and the books in front of you and my utterance and your utterance. (Marschall 2026, 268)

Davidson’s terse reply—“Well I don’t know what you mean by physical account”—shows how different his orientation is to the problems under discussion than the other participants, a difference I have elaborated upon elsewhere (Pearson 2011). But I think that the point upon which Quine and Davidson agree – that presentation predates representation – might be enough to rebuff the problem posed by indirect causal chains in ROBOT and DISTORTION. The mistake was to too quickly allow that the envatted brain in ROBOT was talking about the outside world. Later in the conversation, Davidson reflects upon his doubts about traditional empiricism:

I’m suspicious of the whole floor of information reaching us. I see this as a source of trouble for the empirical tradition going way way back. The idea of little messengers coming in from the outside world and telling us something, and then the skeptic says, “yes but how do you know they aren’t lying?” (Marschall 2026, 366)

Triangulation is meant to be an advance because the causal chains stemming from objects in the world to triangulators are guileless. Whatever the actual common cause is between triangulators will be what their talk and thought is about. Yet opening the door to indirect causation mediated by the person with whom one is triangulating reintroduces the problem Davidson sought to avoid. For like the envatted brain in DISTORTION, if the other person deceptively warps the causal chains that stem to us from the shared world, there is nothing to prevent us from having largely false beliefs about the world.

Instead, what I think Davidson should rule is that for genuine triangulation to occur—the sort necessary for what we call talk and thought—the causal chains that stream to each triangulator may not go through the other. In both ROBOT and DISTORTION, the causal chains allowing conversation between the envatted brain and those outside the vat must therefore be inside the computer’s simulation. HAL is in causal contact with the envatted brain via the electronic signals it sends, and is in causal contact with those outside the vat via the way its signals cause DATA and LORE to move about the world. The objects within HAL’s simulation are what the envatted brain is presented with, and so are what Betty and Elena should interpret the brain as talking about. Given the setup in ROBOT, it is true that Betty can use what the brain says about its simulated world to navigate the real world, since the information HAL sends DATA is an authentic representation of reality. Perhaps the envatted brain could be persuaded that its speech, in addition to expressing truths about the world it is presented, represents truths about the world beyond. But that is not to say that the brain has access to the real world.

This approach has the virtue of retaining the Davidsonian idea that envatted brains in ROBOT and DISTORTION have largely true beliefs. However, it rules out the idea that Betty and the envatted brain in ROBOT are actually conversing about the world outside the vat. An alternative approach that someone (though not me) might advocate is instead to hold on to the possibility of indirect causation and live with skepticism. The envatted brain in ROBOT does have largely true beliefs about the world outside its vat. The envatted brain in DISTORTION has largely false beliefs about the world outside its vat. This has the consequence that an envatted brain has no way of determining if it is in a ROBOT or DISTORTION scenario, and so could consistently worry that it was massively deceived about the world. Correspondingly, an anxious skeptic might worry that they are such a massively deceived envatted brain. I take these considerations to tell against the approach, but perhaps others would find it appealing. Indeed, perhaps Davidson himself would. For he once wrote that a skeptic asking for knowledge that he had knowledge was “demanding more of knowledge than is justified (Davidson 1999a, 162). So perhaps the envatted brain should rest content with some, to my mind, rather cold comfort, in knowing that it could have knowledge of the world outside of its vat.

4 Conclusion

Even though I have argued that Davidson’s theory of triangulation cannot sustain the idea of triangulating with an envatted brain via indirect causal chains in a way that keeps skepticism at bay, reflecting on some of the science fiction cases discussed at the conference helps to capture what distinguishes Davidson’s antiskeptical position from Putnam’s. What Davidson does, and Putnam does not, is to provide an account of language and thought that emphasizes the connection that each of us must have to others, even if one does not have an account of the world or the nature of that connection exactly right. When the conference participants accuse Davidson of being an idealist, he replies “[We can’t be] massively wrong. Yes. If that’s idealism I’m all for it. I didn’t know that was idealism” (Marschall 2026, 228). But later he attenuates this position: “The idealist can answer [skepticism] without bringing in an observer … I don’t assign any meaning to words that aren’t interpretable by another. And that requires common causes” (Marschall 2026, 255, 257). And to a certain sort of solipsistic anxiety that someone like Vatol might have, the fear that one is wholly alone, Davidsonian triangulation may provide a distinct source of peace.

Acknowledgements

I am grateful to Andreas Føllesdal and Benjamin Marschall for locating and editing the transcript of the conversation, and for their invitation to discuss its significance in this special issue. This research was supported by a grant from the Center for Advancement of Research and Scholarship at Bridgewater State University.

Amoretti, M. Cristina. 2011. “Triangulation Between Externalism and Internalism.” In Triangulation: From an Epistemological Point of View, edited by M. Cristina Amoretti and Gerhard Preyer, 47–68. Frankfurt: Ontos Verlag.
Amoretti, M. Cristina, and Gerhard Preyer, eds. 2011. Triangulation: From an Epistemological Point of View. Frankfurt: Ontos Verlag.
Bernecker, Sven. 2013. Triangular Externalism. Edited by Ernie Lepore and Kirk Ludwig. Oxford: Wiley-Blackwell.
Borradori, Giovanna. 1994. The American Philosopher. Translated by Rosanna Crocitto. Chicago: Chicago University Press.
Bridges, Jason. 2006. “Davidson’s Transcendental Externalism.” Philosophy and Phenomenological Research 73 (2): 290–315.
Davidson, Donald. 1978. “What Metaphors Mean.” Critical Inquiry 5: 31–47. Reprinted in Davidson (2001c), pp. 245-64.
———. 1982. “Rational Animals.” Dialectica 36: 318–27. Reprinted in Davidson (2001d), pp. 95-106.
———. 1983. A Coherence Theory of Truth and Knowledge. Reprinted in Davidson (2001d), pp. 137–53.
———. 1987. “Afterthoughts.” In Reading Rorty, edited by Alan Malachowski, 134–37. Cambridge: Blackwell. Reprinted in Davidson (2001d), pp. 154–57.
———. 1990. “Meaning, Truth and Evidence.” In Perspectives on Quine, edited by R. Barrett and R. Gibson, 47--62. Oxford: Blackwell. Reprinted in Davidson (2005).
———. 1991a. “Epistemology Externalized.” Dialectica 45 (2/3): 191–202. Reprinted in Davidson (2001d), pp. 193-204.
———. 1991b. “Three Varieties of Knowledge.” In A.J. Ayer Memorial Essays: Royal Institute of Philosophy Supplement, edited by A. Phillips Griffiths, vol. 30, 30:153–66. Cambridge: Cambridge University Press. Reprinted in Davidson (2001d), pp. 205-20.
———. 1992. “The Second Person.” Midwest Studies in Philosophy 17 (1): 255–67. Reprinted in Davidson (2001d), pp. 107-22.
———. 1997. “Seeing Through Language.” Royal Institute of Philosophy Supplement 42: 15–27. Reprinted in Davidson (2005), pp. 127-42.
———. 1999a. “Reply to Barry Stroud.” In The Philosophy of Donald Davidson, edited by L. E. In Hahn, 162–66. La Salle, IL: Open Court.
———. 1999b. “Reply to Thomas Nagel.” In The Philosophy of Donald Davidson, edited by L. E. In Hahn, 207–09. La Salle, IL: Open Court.
———. 1999c. “The Emergence of Thought.” Erkenntnis 51 (1): 7–17. Reprinted in Davidson (2001d), 123-34.
———. 2001a. “Comments on the Karlovy Vary Papers.” In Interpreting Davidson, edited by Peter Kotatko, Peter Pagin, and Gabriel Segal, 285–308. Stanford: CSLI Press.
———. 2001b. “Externalisms.” In Interpreting Davidson, edited by Peter Kotatko, Peter Pagin, and Gabriel Segal, 1–16. Stanford: CSLI Press.
———. 2001c. Inquiries into Truth and Interpretation. Oxford: Oxford University Press.
———. 2001d. Subjective, Intersubjective, Objective. Oxford: Oxford University Press.
———. 2005. Truth, Language, and History. Oxford: Clarendon Press.
Ellis, Jonathan. 2011. “Stroud’s Proposal for Removing the Threat of Skepticism.” In The Possibility of Philosophical Understanding: Reflection on the Thought of Barry Stroud, edited by W. Wong, N. Kolodny, and J. Bridges, 86–123. Oxford: Oxford University Press.
Føllesdal, Dagfinn. 1999. “Triangulation.” In The Philosophy of Donald Davidson, edited by L. E. Hahn, 719–28. La Salle, IL: Open Court.
Forbes, Graeme. 1999. “Realism and Skepticism: Brains in a Vat Revisited.” In Skepticism: A Contemporary Reader, edited by Keith DeRose and Ted A. Warfield, 61–75.
Genova, A. C. 1999. “The Very Idea of Massive Truth.” In The Philosophy of Donald Davidson, edited by L. E. Hahn, 167–91. La Salle, IL: Open Court.
Glock, Hans. 2003. Quine and Davidson on Language, Thought and Reality. Cambridge: Cambridge University Press.
Goldberg, Nathaniel. 2022. “The Systematicity of Davidson’s Anti-Skeptical Arguments.” Topoi 42 (1): 1–13.
Hahn, L. E., ed. 1999. The Philosophy of Donald Davidson. La Salle, IL: Open Court.
Hirsch, Eli. 2018. Radical Skepticism and the Shadow of Doubt: A Philosophical Dialogue. London and New York: Bloomsbury.
Kotatko, Peter, Peter Pagin, and Gabriel Segal, eds. 2001. Interpreting Davidson. Stanford: CSLI Press.
Lepore, Ernie, and Kirk Ludwig. 2005. Donald Davidson: Meaning, Truth, Language, and Reality. Oxford: Oxford University Press.
———, eds. 2013. A Companion to Donald Davidson. Oxford: Wiley-Blackwell.
Malpas, Jeff. 2011. “Triangulation and Philosophy: A Davidsonian Landscape.” In Triangulation: From an Epistemological Point of View, edited by M. Cristina Amoretti and Gerhard Preyer, 257–80. Frankfurt: Ontos Verlag.
Marschall, Benjamin, ed. 2026. “The 1986 Stanford Conversation Between Quine, Davidson, Dreben, and Føllesdal.” Journal for the History of Analytical Philosophy 16 (2): 13–396.
McDowell, John. 1994. Mind and World. Cambridge: Harvard University Press.
McGinn, Colin. 1986. “Radical Interpretation and Epistemology.” In Truth and Interpretation: Perspectives on the Philosophy of Donald Davidson, edited by Ernie Lepore, 356–68. Oxford: Blackwell.
Mead, George Herbert. 1934. Mind, Self, & Society: From the Standpoint of a Social Behaviorist. Edited by Charles Morris. Chicago: Chicago University Press.
Morris, Charles. 1937. Logical Positivism, Pragmatism, and Scientific Empiricism. Paris: Hermann.
Myers, Robert H., and Claudine Verheggen. 2016. Donald Davidson’s Triangulation Argument: A Philosophical Inquiry. London and New York: Routledge.
Nagel, Thomas. 1999. “Davidson’s New Cogito.” In The Philosophy of Donald Davidson, edited by L. E. Hahn, 195–206. La Salle, IL: Open Court.
Pagin, Peter. 2001. “Semantic Triangulation.” In Interpreting Davidson, edited by Peter Kotatko, Peter Pagin, and Gabriel Segal, 199–212. Stanford: CSLI Press.
Pearson, James. 2011. “Distinguishing W. V. Quine and Donald Davidson.” Journal for the History of Analytical Philosophy 1 (1): 1–22.
———. 2023. “Objectivity Socialized.” In The Philosophical Project of Carnap and Quine, edited by Sean Morris, 92–113. Cambridge: Cambridge University Press.
Pritchard, Duncan. 2013. Davidson and Radical Skepticism. Edited by Ernie Lepore and Kirk Ludwig. Oxford: Wiley-Blackwell.
Putnam, Hilary. 1981. Reason, Truth, and History. Cambridge: Cambridge University Press.
Quine, W. V. 1992. Pursuit of Truth. Revised Edition. Cambridge, MA: Harvard University Press.
Ramberg, Bjorn. 2001. “What Davidson Said to the Skeptic.” In Interpreting Davidson, edited by Peter Kotatko, Peter Pagin, and Gabriel Segal, 213–36. Stanford: CSLI Press.
Stroud, Barry. 1999. “Radical Interpretation and Philosophical Scepticism.” In The Philosophy of Donald Davidson, edited by L. E. Hahn, 139–61. La Salle, IL: Open Court.
Wittgenstein, Ludwig. 1953. Philosophical Investigations. Translated by G. E. M. Anscombe. Oxford: Basil Blackwell.
Yalowitz, Steven. 1999. “Davidson’s Social Externalism.” Philosophia 27: 99–136.

  1. Some critics distinguish the argument that only triangulators can possess certain concepts needed for thought—such as objectivity, belief, or error—from the argument that only triangulation allows for the object-directedness of thought (see, for instance Lepore and Ludwig 2005; and Bridges 2006). Others insist that these must be understood as aspects of the same argument (see Myers and Verheggen 2016).↩︎

  2. See, for instance, Davidson (1992, 118–19).↩︎

  3. As we shall see below, our interpretation of this caveat will be crucial to understanding the antiskeptical force of triangulation.↩︎

  4. If we choose to credit animals’ various cognitive achievements by attributing our concepts of belief, thought, and intention to them, I think that Davidson would say that we are speaking metaphorically. He would count such descriptions as literally false but concede that they may allow us to articulate a striking feature of our world. As he puts it, “metaphors often make us notice aspects of things we did not notice before; no doubt they bring surprising analogies and similarities to our attention; they do provide a kind of lens or lattice…through which we view the relevant phenomena” (Davidson 1978, 261). Nevertheless, we should be careful not to confuse metaphor with literal truth. Attributing “beliefs” to animals is on a continuum with metaphors used in naïve science: “someone might easily have no better or alternative way of explaining the movements of a heat seeking missile than to suppose the missile wanted to destroy an airplane and believed it could by moving in the way it was observed to move” (Davidson 1982, 101).↩︎

  5. We may creatively describe cases where a solitary being has misleading experiences of the world, cases in which we would describe them as making mistakes. (See, e.g., Lepore and Ludwig 2005, 402) But their experience does not suffice for them to grasp the concept of error and so to have what we call beliefs. For nothing about their experience explains how they could become obliged to correct their mistakes, since there is nothing about their experience to distinguish correcting their view from merely changing their view.↩︎

  6. We might, for example, describe a startled cat as surprised by its experience, without attributing second order beliefs to it.↩︎

  7. For variations on this objection, see Yalowitz (1999), Pagin (2001), and Bernecker (2013).↩︎

  8. Jeff Malpas applauds triangulation’s circularity as a principled rejection of “what philosophy often seems to demand, namely, some privileged level of analysis on the basis of which all else can be extrapolated, explained, or deduced” (Malpas 2011, 266).↩︎

  9. The “problem of emergence” identified by Mead’s student and editor, Charles Morris, also pertains to the different sorts of triangulation Davidson distinguishes (Morris 1937, 34–35). If Mead is right that there is a sharp difference between the thought of humans and the thought of animals, Morris worries, how can we use language whose content refers to our self-conscious experience to describe the experience of the pre-social beings from whom we emerged? As Davidson puts it, “There is a stage at which there is no thought followed by a subsequent stage at which there is thought … What we lack is a satisfactory vocabulary for describing the intermediate steps … You can deceive yourself into thinking that the child is talking if it makes sounds which, if used by a genuine language user, would have a definite meaning. (It is even possible to do this with chimpanzees.) But words, like thoughts, have a familiar meaning, a propositional content, only if they occur in a rich context, for such a context is required to give the words or thoughts a location and a meaningful function” (Davidson 1997, 127).↩︎

  10. For further discussion of Mead’s call to reorient epistemology, see Pearson (2023).↩︎

  11. At least one early critic sought to argue that Davidson’s philosophy was mistaken because it had antiskeptical consequences. “How could epistemologists have been so wrong about what is necessary to defeat the sceptic?” Colin McGinn wondered, astonished that “Davidson regarded this as a virtue of his account” (McGinn 1986, 359, original emphasis).↩︎

  12. Nathaniel Goldberg lucidly distinguishes three interlocking antiskeptical arguments in Davidson’s oeuvre: one targeting the idea that beliefs are true only relative to conceptual schemes, another targeting the idea that beliefs could generally be false, and a third targeting the idea that the world and other minds might not exist. It is the third of these arguments that arises from triangulation. Goldberg distinguishes two “versions” of this argument, one focusing upon the concept of objectivity and a second on the content of beliefs (Goldberg 2022, 8). But I agree with Myers and Verheggen (2016) that Davidson’s reflections on triangulation are made most plausible if they are read in concert.↩︎

  13. For useful discussion, see Ellis (2011).↩︎

  14. Davidson consistently urges the others to take his underlying criticisms of Quine seriously: “Science fiction cases are easy to cook up and are dramatic, but we don’t need them” (Marschall 2026, 190); “There’s nothing sort of science fictiony about this” (Marschall 2026, 209); “I’ve got all kinds of reasons just in ordinary situations without any science fiction whatsoever to think …” (Marschall 2026, 375–76).↩︎

  15. Later in print, Davidson echoes the attitude of the other conference participants: “I have a general distrust of thought experiments that pretend to reveal what we say under conditions that never in fact arise” (Davidson 1990, 197).↩︎

  16. KIDNAP is inspired by remarks from Davidson and Føllesdal (Marschall 2026, 223, 226).↩︎

  17. M. Cristina Amoretti suggests that since the beliefs formed by a newly envatted brain are caused by its current environment a Davidsonian may eventually interpret it as having true beliefs whose content concerns that environment—provided that the diabolical scientist does not keep cycling the brain into the vat and out again in such a way that it never establishes stable causal links (Amoretti 2011, 65). Concluding that Davidson at best has the resources to argue that our beliefs “cannot always have been” totally false, not that they “can never be” totally false, she thinks we may nevertheless have a prima facie reason to believe that our current, largely coherent set of beliefs is probably massively true. But I cannot see what grounds this last claim about probability. Why should we judge it more likely that our current beliefs are massively true than massively false, if we have accepted that the newly envatted brain’s beliefs are just as coherent as our own?↩︎

  18. SUPERCOMPUTER is inspired by Putnam (1981, 12) and Davidson’s remarks at the conference (Marschall 2026, 223–25).↩︎

  19. Davidson disagrees. At the conference, he remarks approvingly “as Quine has said again and again, theory of knowledge and theory of meaning have the same starting point” (Marschall 2026, 368). In externalizing semantics but not epistemology, Davidson thinks Quine has missed his own lesson (Davidson 1991a, 194). For a more detailed discussion of the differences between their late views, see Pearson (2011).↩︎

  20. SOLIPSISM is inspired by Davidson’s remarks at the conference (Marschall 2026, 357).↩︎

  21. Might the persistent skeptic object that the simulated people need only seem to be dynamic? Perhaps the program running on HAL just happens to stimulate the brain in just the right way(s) at just the right time(s), so that while it seems that the simulated people in the environment are responding to the brain’s behavior, this is a pure coincidence. But as the conference participants would be quick to note, the more we are forced to appeal to science fiction, the less of philosophical interest remains. Adopting Davidson’s strategy with KIDNAP, I would isolate skeptical scenarios that rely upon this sort of cosmic coincidence and simply acknowledge that there is no answer to them.↩︎

  22. ROBOT is inspired by Davidson’s, Dreben’s, and Føllesdal’s remarks at the conference (Marschall 2026, 256, 309–10, 378).↩︎

  23. “Before the turn of the century,” Dreben enthuses, technology will have advanced enough so that “some blind people will have” the sort of ability Davidson describes (Marschall 2026, 378). “But notice,” Davidson replies, “we’ve got such things now … A little radar running into a cane, it prints out stuff on your chest or on your back, I forget which. Patterns of stimulation are completely different, but I certainly say that I can talk to the blind guy about the same thing because we’re both responding to the same thing” (Marschall 2026, 378).↩︎

  24. DISTORTION is inspired by Føllesdal’s remarks at the conference (Marschall 2026, 249–50).↩︎

  25. Here I am assuming that the brain’s utterance of “roughly spherical” has its regular English meaning. Had Elena trained the brain that banana-shaped objects are “roughly spherical,” a critic might object that the brain is speaking a different language where “roughly spherical” means “banana-shaped,” and so, that its belief is true after all. But there is no reason why we must make this assumption: after initially accurate training with shapes and colors, Elena could introduce (and systematically lie about the true nature of) fruits, and much more besides.↩︎

  26. Doubtless picture books quickly become part of this story for the majority of language learners, but we should have no reason to interpret a person solely familiar with picture books to be talking about what the pictures depict as opposed to the pictures themselves.↩︎

Copyright © 2026 James Pearson