Our paper contributes to the rapidly expanding body of experimental research on transformative decision making, and in the process, marks out a novel empirical interpretation for assessments of subjective value in transformative contexts. We start with a discussion of the role of subjective value in transformative decisions, and then critique extant experimental work that explores this role, with special attention to Reuter and Messerli (2018). We argue that current empirical treatments miss a crucial feature of practical deliberation manifesting across a variety of everyday decisions: often, people attach more weight to decision criteria that they can know—a phenomenon known as the “evaluability bias.” In transformative contexts, if people cannot know the subjective value of an option, they are unlikely to attach it much weight. Despite this, people may care very much about such value. We then use this point to develop and present new empirical results that, in addition to supporting our concerns about evaluability bias, support the hypothesis that people care about subjective value. Our work enriches the current philosophical understanding of transformative decisions and helps to frame the emerging experimental paradigm for the empirical dimensions of the debate.

In what sense does a large language model have knowledge? The answer to this question extends beyond the capabilities of a particular AI system, and challenges our assumptions about the nature of knowledge and intelligence. We answer by granting LLMs "instrumental knowledge"; knowledge defined by a certain set of abilities. We then ask how such knowledge is related to the more ordinary, "worldly" knowledge exhibited by human agents, and explore this in terms of the degree to which instrumental knowledge can be said to incorporate the structured world models of cognitive science. We discuss ways LLMs could recover degrees of worldly knowledge, and suggest such recovery will be governed by an implicit, resource-rational tradeoff between world models and task demands.

The discovery of a new kind of experience can teach an agent what that kind of experience is like. Such a discovery can be epistemically transformative, teaching an agent something they could not have learned without having that kind of experience. However, learning something new does not always require new experience. In some cases, an agent can merely expand their existing knowledge using, e.g., inference or imagination that draws on prior knowledge. We present a computational framework, grounded in the language of partially observable Markov Decision Processes (POMDPs), to formalize this distinction. We propose that epistemically transformative experiences leave a measurable "signature" distinguishing them from experiences that are not epistemically transformative. For epistemically transformative experiences, learning in a new environment may be comparable to "learning from scratch" (since prior knowledge has become obsolete). In contrast, for experiences that are not transformative, learning in a new environment can be facilitated by prior knowledge of that same kind (since new knowledge can be built upon the old). We demonstrate this in a synthetic experiment inspired by Edwin Abbott's Flatland, where an agent learns to navigate a 2D world and is subsequently transferred either to a 3D world (epistemically transformative change) or to an expanded 2D world (epistemically non-transformative change). Beyond the contribution to understanding epistemic change, our work shows how tools in computational cognitive science can formalize and evaluate philosophical intuitions in new ways.

Some experiences transform us in profound ways. Many people, for instance, are transformed by becoming a parent. How does this occur? Is parenthood like a massive hurricane, which shapes a wide landscape directly and all at once by overwhelming force? Or could experiences like parenthood transform us the way a spark transforms the landscape—as a small force that only directly touches one dry leaf, but thereby sets off a chain reaction? We explore the possibility and implications of spark-like personal transformative experiences––those in which a relatively small, surgical adjustment of one aspect of our preferences has an extended ripple effect, rationally compelling the revision of many other aspects of our preferences––in the context of current models of the notion of ‘desire’ in the cognitive sciences, focusing especially on theories of value-guided decision-making.

Choosing brings change. A major life choice, a “big decision,” as Edna Ullmann-Margalit termed it, brings the possibility of changing who you are. Richard Pettigrew’s important new book, Choosing for Changing Selves, explores theories of decision-making for choices that change us, with a particular focus on life-changing decisions.Life-changing decisions are decision cases where a persisting agent, through their choice, creates a new self, through replacing their values—that is, replacing that self’s utility function. (Take the persisting agent to be constituted by a series of appropriately related selves, and selves to be defined by their values, and by extension by their utility functions.) Such cases are of deep philosophical and practical interest, involving questions about knowledge, evidence, and experiential value, and concerning real world choices such as choosing to have a child, determining one’s future medical care, or getting a divorce (Bykvist 2006; Ullmann-Margalit 2006; Paul 2014).The central focus of the book is on cases where an agent, through her life-changing choice, replaces her utility function. If your values change as the result of your choice, which values govern your action? Your values at the time of your choice (your ex ante values)? Or your values at the time of the outcome of your choice (your ex post values)?Spelled out in terms of the metaphysics of selves: at t1, the agent is realized by her ex ante self, whose values are defined with utility function U1. This self deliberates, and chooses to act in a way that leads to changes in U1. By changing herself in this way, she realizes her ex post self at t2, a new self with new values defined by utility function U2. If these value changes go “all the way down,” there is no way to choose that is consistent with both U1 and U2. Take a person who, at t1, highly values being child-free, as defined by her utility function U1, and does not value being a parent. However, if she becomes a parent at t2, she will highly value that state (using utility function U2). Traditional decision theory requires that, to choose rationally, she must choose in accordance with her values. In such a case, which self’s values should determine her choice?As Pettigrew demonstrates with clear, elegant prose and mathematical precision, given a small set of reasonable assumptions, making these choices rationally requires one to modify orthodox expected utility theory. Blending formal epistemology with the metaphysics of selves and persisting agents, he proposes a new theory of rational choice for persisting agents.His solution, in brief, is that one should choose in accordance with a weighted general value function (UG) that is the amalgamation of the local functions U1, U2, …, Un of one’s n selves—that is, an amalgamation of the value functions of all the past, present, and future selves that compose the life of the agent. Inspired by the work of Derek Parfit and others, he develops and defends an account of UG as a weighted average of the local utility functions of the different selves that compose the persisting agent over time, where the weights are determined by sacrifices made by local selves, psychological connectedness, and the degree of shared values.A key move in Pettigrew’s argument for UG is to treat the rationality of choosing for changing selves as a judgment aggregation problem, where the credences and utilities of distinct selves are aggregated. This move is modeled explicitly on the literature for collective decision-making. Chapter 5 nicely develops the way the weak reflection principle applies in cases of credal (or value) change, showing how principles for credences about objective chances depend on the structure of the situation. In particular, in certain types of collective knowledge and peer disagreement contexts (Titelbaum and Kopec 2009), rationality can dictate how a person should incorporate facts about the credences of other persons into their own credences.Drawing heavily on an analogy between a person’s relations to other people and one’s current self’s relations to one’s past and future selves, Pettigrew then claims that facts about the utilities of one’s different selves should be taken into account in contexts of self-change, using collective norms to calculate the local utilities of past, present, and future selves and incorporate them into a general utility function.1 The result is Pettigrew’s solution to the problem of choosing for changing selves: credences and utilities of distinct selves (on analogy to cases of collective decision-making involving distinct persons) are weighted and aggregated to determine a general value function (UG), and when an agent makes a life-changing decision, they should maximize their utilities in accordance with UG.I found this systematic approach enlightening and fascinating. The mathematical treatment of utility functions as applied to the question of self-change, set in the context of contemporary formal epistemology, is especially valuable. The book is packed with interesting ideas about how to treat utilities for persisting agents, and does an especially good job of weaving these insights into extant discussions of credal change. However, I am not convinced by Pettigrew’s solution. Space is limited, so I will only discuss two objections.First, why think that the relations and requirements for how we treat other people (“collective norms”) should govern the way we treat our other selves? Pettigrew takes the premise to be obviously true. I am skeptical. In fact, there is excellent evidence that, in important contexts, we treat our own selves more harshly than we treat other people (Crockett et al. 2014).A related question concerns the weights we assign to selves. Pettigrew defends a “principle of minimal mutilation,” placing special emphasis on the sacrifices one’s past selves have made. But why should my past selves be treated with such deference? I see no reason now to place any weight at all on my five-year-old self’s dislike for well-aged tawny port, or for that matter, on my twenty-year-old self’s facile romantic attractions, no matter how many sacrifices those selves made for me. I want the freedom to dispense with my past mistakes, to start anew, to throw off the shackles of my former selves.The second objection runs deeper. Pettigrew’s approach requires us to make meaningful intraself utility comparisons. The utility function (U1) of the ex ante self must be comparable to the utility function (U2) of the ex post self, or if U1 and U2 are incompatible, we must be able to scale them in order for utility changes to be meaningful. (By analogy, if I tell you that it was 25 degrees outside at t1, but that it is now 35 degrees at t2, this is meaningful only if (1) you know the temperature scales I am using, and (2) you can convert values in one to values in another. Warming from 25°F to 35°F is not the same as warming from 25°F to 35°C.)Pettigrew addresses this problem by drawing on extant models of interself utility comparisons, and develops a very beautiful solution for intraself comparisons: the “matching intervals solution.” This method allows one to set the zero and unit of the scale for different utility functions of one’s different selves in order to ensure that the functions are comparable. Assuming (!) for the sake of discussion that all relevant incompatibilities can be resolved by scaling, we can make meaningful comparisons between our current and future (or counterfactual) selves as long as we can compare the right intervals. The solution requires an individual to be able to determine and compare differences between their current utilities for pairs of outcomes (o1, o2) to differences between a current utility for o2 and their future (or counterfactual) utility for o2.Unfortunately, the problem of incomparable value functions resurfaces when Pettigrew considers transformative choices. An agent makes a transformative choice when their ex ante self, with utility function U1, chooses to act in a way that replaces their current utility function, in effect, creating an ex post self at t2 with a new (incompatible) utility function U2.When you choose to transform, your choice will create a new self, with a new utility function. To evaluate the desirability of this choice, you must compare the utilities assigned to the outcomes by your new self with the utilities assigned to the outcomes by your old self. For the comparison to be meaningful, the utilities of these selves must be comparable. (For example, the way you’d value parenthood when you’re child-free [using U1] is different from the way you’d value parenthood as a parent [using U2]. To meaningfully interpret this difference, U1 and U2 must be comparable.)The intuitive way we try to compare utilities across selves is by prospectively assessing the utilities of our future selves. If you know what it would be like to be some future self, you can imagine yourself this way, empathically assess your utilities (and scale if needed), and meaningfully compare this utility to your current utility.The trouble is that, in transformative cases, we cannot simply imaginatively evolve ourselves forward in order to empathize with our future selves in this way. We lack the necessary prospective abilities. Pettigrew’s solution to this problem is to substitute other-based testimony about utility changes, drawn from social-scientific research on people who are relevantly like you.However, if we are to be justified in relying on other-based testimony about utility change when we are considering a class of cases involving the possibility of incomparable utility functions across our different selves, we must know that the reported utilities are comparable. Unfortunately, the problem of incomparable value functions infects the interpretation of existing empirical results (Paul and Healy 2018).In brief: in randomized controlled trials (RCTs), utility changes are measured by comparisons (of average results) between a treatment group and a control group. In the cases of interest, over a fixed period of time, members of the treatment group undergo transformative experiences. Members of the control group do not. All else is held fixed. After the experience, utility changes of members of the treatment group are measured and compared to utility changes (if any) of members of the control group. On the assumption that the relevant utility comparisons are meaningful, any differences in (average) utility between the two groups can be seen as a measure of the impact of the transformative experience.But we cannot assume meaningfulness. A version of the problem of intrapersonal incomparability arises for the selves of the treatment group before and after their treatment (before and after the transformative experience). In these studies, the utilities of treated subjects are compared at t1, before (ex ante) the “treatment” (the transformation) and at t2, after the treatment (ex post). By definition, transformation involves utility function replacement. If the subjects’ ex post utility functions are not comparable to their ex ante utility functions, comparisons of their ex ante utilities with their ex post utilities will not be meaningful.You might hope that the problem can surmounted, for the important comparison in these studies is at t2, between the ex post utilities of the treatment group and the ex post utilities of the control group. But the problem remains. For an RCT, the meaningfulness of the ex post comparison between the control group and the treatment group relies on matching the two groups before the treatment is applied, and then assuming this match carries forward throughout the experimental procedure. (The need for such a match stems from empirical constraints concerning the use of counterfactuals, and derives from what social scientists term “the fundamental identity problem.”) The change in utility functions destroys the match, destroying inferences relying on comparability.In brief, ex post comparability between the utilities of treatment and control groups can fail, and for precisely the same reason as comparability between the utilities of one’s current and counterfactual selves can fail.There is much more to engage with in Pettigrew’s absorbing book than I can treat in this short review. My objections notwithstanding, there is no question that it makes a major contribution to the literature. It is filled with insightful ideas and new connections. It breaks new ground by developing a systematic mathematical theory of the role and relationships between the utility functions of selves (the existing literature focuses almost solely on credence functions), and by applying this mathematical approach to the range of metaphysical and epistemological issues that arise when choosing for changing selves. It is a model of clarity and precision. As such, Choosing for Changing Selves is essential reading for anyone interested in decision theory, practical reasoning, agent persistence and change, temporal discounting, and personal identity.

We distinguish two types of cases that have potential to generate quasi-cyclical preferences: self-involving choices where an agent oscillates between first- and third-person perspectives that conflict regarding their life-changing implications, and self-serving choices where frame-based reasoning can be “first-personally rational” yet “third-personally irrational.” We argue that the distinction between these types of cases deserves more attention in Bermúdez's account.

Abstract A commitment to truth requires that you are open to receiving new evidence, even if that evidence contradicts your current beliefs. You should be open to changing your mind. However, this truism gives rise to the paradox of empathy. The paradox arises with the possibility of mental corruption through transformative change, and has consequences for how we should understand tolerance, disagreement, and the ability to have an open mind. I close with a discussion of how understanding this paradox provides a new explanation for a certain kind of standoff between the believer and the skeptic with regard to religious belief.

Universities aspire to offer students a transformative experience, but rarely spell out the nature of this transformation. In this essay, L. A. Paul and John Quiggin frame the successful university education as transformative in the philosophical sense. They explain the way that a successful college education can be understood as generating an individual conceptual revolution, and thus, as a transformative experience. Education can create epistemically transformative change through the process of developing critical thinking skills, leading to conceptual replacement and the discovery of new intellectual frameworks. This epistemic transformation, if deep enough, scales up into a personal transformation. After explicating the nature and structure of transformative education, Paul and Quiggin show how understanding transformation in terms of personal change and awareness of unawareness clarifies the debate over its value.

Transformative experiences are experiences that change the way you understand the world, and in the process, they change who you are. As such, they are experiences that are both epistemically and personally transformative. The epistemic transformation involves a discovery: until you undergo the experience, you don’t know what it will be like to be your future, transformed self. The personal transformation involves the way the experience changes the kind of person you are (or more colloquially, “changes who you are”). Who you’ll become is different, in some important and salient way, from who you are now. In my (Paul 2014), I develop an account of the conceptual and psychological structure of transformative experience. Using this account, I argue that the self-changes such experiences involve, combined with the epistemic limitations we face with regard to what we can first-personally know before the transformation, raises distinctive problems for decisionmaking in transformative contexts. On my approach, a persisting person is taken to be a temporally extended entity composed of a series of continuous, appropriately related selves. A person who persists through a transformative experience is realized by one kind of self before the transformation and by a different kind of self after the transformation: the new, replacement self, is created in response to the transformative experience. Examples of transformative experiences include fighting in a war, having one’s first child, becoming sighted after a life of blindness, getting a cochlear implant, losing a loved one, getting divorced, and having a religious conversion. There are two especially important features to note about the structure of these types of experiences. First, because it is unlike other kinds of experiences they have had, in an important and distinctive sense, a person cannot imagine what the transformative experience will be like for them before they undergo it. Second, because it will change the kind of person they are, some central and important elements of what they care about will change. As Walsh (2020) argues, cases of dementia can be instances of cognitively transformative experiences: they can be cases where a person has an epistemic transformation that creates a personal transformation. In dementia, the epistemic transformation involves a radical change in an individual’s cognitive capacities, changing, in a deep and salient way, the nature of their lived experience of their relationships, their cognitive abilities, and their sense of the world around them. These changes lead to personal transformation: changes in some of their core preferences, for example, their preferences regarding lifeprolonging treatment in medical contexts. This type of cognitively transformative experience exhibits the classic structure of transformation, where the self that exists before the transformation (the “ex ante” self) is psychologically incommensurable with the self that is created by the transformation (the “ex post” self). The incommensurability stems, in the first instance, from a radical change in the nature and character of the individual’s lived experience, which results in one or more core preference changes. Importantly, there is no higher-order preference structure that is consistent across the changes in self: there is no way to make the preferences of the ex ante self consistent with the ex post self. (For discussion, see Paul 2015a, 2015b; Pettigrew 2015; Paul and Quiggin 2018, and Pettigrew 2019.) For example, Mrs. Black (ex ante) prefers not to receive life-prolonging medical treatment if she can longer recognize friends and family, while Mrs. Black (ex post) prefers the medical treatment, as she is happy and content with her current state, despite her inability to recognize friends and family members. In the case as described, there is no higher-order preference that can resolve this conflict. Before the transformation, she prefers not to receive the treatment. Afterwards, she does. This raises the pressing question: which temporal stage (which self) of Mrs. Black should decide the question of whether she is to receive treatment? In my (Paul 2014), I discussed these issues in the context of concerns about disability, and suggested that cases

Abstract In this chapter, Jeff Sebo and L.A. Paul investigate the phenomenon of experiences that transform the experiencer, either epistemically, personally, or both. The possibility of such experiences, Sebo and Paul argue, frequently complicates the practice of rational decision-making. First, in transformative cases in which your own experience is a relevant part of the outcome to be evaluated, one cannot make well-evidenced predictions of the value of the outcome at the time of decision. Second, in cases in which one foresees that one’s preferences would change following the decision, there are issues about whether rational decision-making should be based only on one’s ex ante preferences, or should also incorporate some element of deference to foreseen future preferences. While these issues arise quite generally, Paul and Sebo suggest that they are especially pressing in the context of effective altruism.

ABSTRACTIn the real world, there can be constraints on rational decision-making: there can be limitations on what I can know and on what you can know. There can also be constraints on my ability to deliberate or on your ability to deliberate. It is useful to know what the norms of rational deliberation should be in ideal contexts, for fully informed agents, in an ideal world. But it is also useful to know what the norms of rational deliberation should be in the actual world, in non-ideal contexts, for imperfectly informed agents, especially for big, life-changing decisions. That is, we want to know how to deliberate as best we can, given the real-world limitations on what we can know, and given real-world limitations on how we are able to deliberate. In this paper, our concern is with the norms of rational deliberation in certain, important, non-ideal contexts, where the reasoning occurs from the agent's first person, subjective point of view. The norms governing the process of deliberation for real people in the sorts of non-ideal contexts we'll consider need to reflect the way that real agents, with an incomplete grasp on the facts and an imperfect ability to deliberate, can be expected to proceed. Our central contention is that framing and exploring the deliberative process from the first person perspective allows us to uncover and explore important, real-world constraints on boundedly rational agents deliberating from the subjective perspective.

Philosophical PerspectivesVolume 31, Issue 1 p. 7-39 Original Article DE SE PREFERENCES AND EMPATHY FOR FUTURE SELVES1 L.A. Paul, L.A. Paul University of North Carolina at Chapel Hill University of St AndrewsSearch for more papers by this author L.A. Paul, L.A. Paul University of North Carolina at Chapel Hill University of St AndrewsSearch for more papers by this author First published: 24 April 2018 https://doi.org/10.1111/phpe.12090Citations: 8Read the full textAboutPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Citing Literature Volume31, Issue1December 2017Pages 7-39 RelatedInformation

Phenomenal character (or feel) is the what‐it's‐likeness of subjective experience. I develop an ontology of phenomenal feel as process. My being in some phenomenal state R is the process of my instantiating R’s neurological correlate. The ontology explains why we have asymmetric epistemic access to phenomenal characters: the ontological ground for the subjective or first‐personal stance is different from the ontological ground for the objective or third‐personal stance. I end by situating my account in debates about physicalism.

I argue that we can understand the de se by employing the subjective mode of presentation or, if one's ontology permits it, by defending an abundant ontology of perspectival personal properties or facts. I do this in the context of a discussion of Cappelen and Dever's recent criticisms of the de se. Then, I discuss the distinctive role of the first personal perspective in discussions about empathy, rational deference, and self-understanding, and develop a way to frame the problem of lacking prospective access to your future self as a problem with your capacity to imaginatively empathize with your (possible) future selves.

Philosophy and Phenomenological ResearchVolume 91, Issue 3 p. 794-813 Book Symposium Transformative Experience: Replies to Pettigrew, Barnes and Campbell L.A. Paul, L.A. Paul University of North Carolina, Chapel HillSearch for more papers by this author L.A. Paul, L.A. Paul University of North Carolina, Chapel HillSearch for more papers by this author First published: 26 October 2015 https://doi.org/10.1111/phpr.12250Citations: 17Read the full textAboutPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Citing Literature Volume91, Issue3November 2015Pages 794-813 RelatedInformation

In "What You Can't Expect When You're Expecting," I argue that, if you don't know what it's like to be a parent, you cannot make this decision rationally at least, not if your decision is based on what you think it would be like for you to become a parent. My argument hinges on the idea that becoming a parent is a trans formative experience. This unique type of experience often transforms people in a deep and personal sense, and in the process, changes their preferences.In section 1, I will explain transformative experience in terms of radical first-personal epistemic and self change. In section 2, I'll explain the notion of subjective value that I use to develop the decision problem. In section 3, I will discuss the way we ordinarily combine our introspective assessments with testimony and evidence. In section 4, I will discuss the problems for rational decision-making. In section 5, I will explore the problem of first-personally transformed future selves. In section 6, I will engage with the main themes and arguments and ideas of the authors of the papers contributed to this volume.

It seems natural to choose whether to have a child by reflecting on what it would be like to actually have a child. I argue that this natural approach fails. If you choose to become a parent, and your choice is based on projections about what you think it would be like for you to have a child, your choice is not rational. If you choose to remain childless, and your choice is based upon projections about what you think it would be like for you to have a child, your choice is not rational. This suggests we should reject our ordinary conception of how to make this life-changing decision, and raises general questions about how to rationally approach important life choices.