What this covers

L.A. Paul discusses her work on transformative experience with Russ Roberts in this EconTalk episode. The conversation centers on a deceptively simple problem: how should we decide whether to undergo experiences that fundamentally change who we are—becoming a parent, religious conversion, disability, or the thought-experiment case of becoming a vampire? Paul argues that these choices resist the standard tools of decision theory because they violate two basic assumptions. First, you cannot know what an experience is like—what philosophers call its epistemic transformation—until you have lived through it, yet most such choices are irreversible. Second, the experience changes your core preferences and identity, so the person deciding and the person who will have decided may want incompatible things, with no principled way to determine which self should take priority.

The discussion ranges across both the theoretical machinery and its real-world consequences. Roberts and Paul examine how standard decision tools break down: the reference class problem makes average evidence about others unreliable for your case, expected utility ignores the full distribution of outcomes and downside risk, and reductionism to a scalar—collapsing multifaceted experiences into single numbers—commits what Paul calls a violence against human reality. The conversation then moves to concrete cases: whether to give a deaf child cochlear implants, the gendered calculus of motherhood for women with established careers, and the cultural shift that has left individuals to author their lives without the traditional guidance of religion and social norms. Throughout, Paul advocates for epistemic humility—acknowledging what we cannot know in advance—while respecting both those who embrace revelation and those who prefer the known.

Sharpest takeaway

L.A. Paul argues that transformative experiences (becoming a parent, religious conversion, disability, the vampire thought experiment) are both epistemically and personally transformative, which makes them resistant to standard expected-utility decision-making because you cannot know what the experience is like or who you will become until after the irreversible choice is made.

  • You cannot assign value to an outcome you cannot experientially understand in advance (epistemic transformation)
  • The choice changes your core preferences, so the self deciding and the self resulting may disagree with no meta-rule to resolve which self matters (act-state dependence)
  • Standard tools — reference classes, average values, imaginative projection — break down precisely in these high-stakes, irreversible cases

The argument · threads4 threads · 28 claims
0.71

Epistemic humility and rules of thumb, combined with cultural narratives where they exist, offer a more honest decision approach than false certainty.

6 pointscentrality 4/5
  • The appropriate response to transformative-experience problems is epistemic humility—recognizing what we can and cannot know and not setting ourselves impossible tasks—and from that stance valuing revelation while respecting those who prefer the known, then building alternative decision models that respect first-person experience rather than quantifying everything from an impersonal quasi-Godlike perspective.

    “I want us to recognize what we can do and what we can't do. What we can know, and what we can't know. Not set ourselves impossible tasks. So, take a stance involving epistemic humility”

  • Because humans, unlike an idealized chess computer, cannot see far down the decision tree, the rational strategy is to adopt rules of thumb (e.g., protect your queen, trade a pawn for a bishop) that give the best chance given limited foresight—a model of bounded, humble decision-making applicable to life choices generally.

    “if you're mediocre, then you can't see that far down the decision tree... the solution is instead to adopt rules to try to make the best decision given what you can see”

  • For some life-changing decisions with incompatible selves—like getting a frontal lobotomy—culture supplies a dominant rule that resolves the choice, but for others like having a child there is no dominant cultural narrative, so the decision is left to the individual, who then faces a genuine decision problem with no rule to follow and inadequate tools to decide.

    “as soon as we leave it to the individual, and if what we're leaving to the individual is the choice of whether or not to undergo a life-changing experience... then, we're confronted with a decision problem--because we don't have any kind of rule to follow”

  • The decision to have a child is highly gendered: in contemporary Western society women with active careers and identities they love often face the choice of whether to pursue this revelation knowing they will likely have to give up significant things they value, a higher-stakes choice with genuine potential negative value that men generally do not face.

    “I think this is a kind of decision, this one in particular is much higher-stakes, often, for women than for men. And it's because that negative value can actually be there.”

  • Social norms constantly change and new situations like climate change or new technology require developing new norms rather than following old ones, and we shouldn't always follow old norms because some of them were problematic, unfair, or cruel.

    “norms are constantly changing. We face new situations as well, like climate change or new technology that requires developing new norms.”

  • We should not assume that someone who, after a terrible accident or as a member of the deaf community, testifies to the value of their new life is confused or experiencing cognitive dissonance, nor that a life with ordinary hearing is more valuable than a flourishing deaf life; such radical assumptions should be resisted.

    “we should never assume that just because someone's had a terrible accident and then testifies to the value of their new life is somehow confused or is experiencing cognitive dissonance or something like that”

0.70

Transformative experiences alter core identity, making prior preferences unreliable for evaluating the choice.

3 pointscentrality 5/5
  • Transformative choices violate the act-state independence axiom because the act of undergoing the experience changes who you are, so the state (your post-choice self and its satisfaction) is dependent on the act rather than independent of it; you cannot tell whether your future satisfaction reflects a pre-existing preference being realized or your preferences simply being replaced, leaving no principled way to decide which self should take priority.

    “the act that you undergo, and the state that you have at the end of it are dependent on each other. Who you are as a person changes; and so there's no straightforward way to make the decision”

  • Transformative choices like becoming a vampire change you in a deep and fundamental way—altering core desires, tastes, and how much you care about others—so even unanimously positive testimony from those who made the choice is hard to evaluate because it reflects the preferences of the changed self, not the self deciding.

    “if their preferences changed radically, then whose preferences matter when you make that decision: the preferences of the vampire, or the preferences of the human that made the decision?”

  • Deciding whether to become a vampire is intractable because the value of the choice turns on what it would be like to live as a vampire, yet that experience can only be known once you become one and the choice is irreversible, so you lack the information needed to make an informed decision and your preferences remain incomplete.

    “if it's the kind of thing that you can only know and understand once you become a vampire, and if it's irreversible--this is a one-shot sort of thing--then you lack the kind of information that you need to make an informed decision”

0.70

Standard decision theory fails by ignoring the irreducible gap between first-person and third-person perspectives on value.

3 pointscentrality 5/5
  • There is an unresolvable split between the first-person perspective (how it feels from inside one's life) and the third-person perspective (how a life looks from outside), and the failure of standard decision theory to recognize and manage these distinct perspectives—sometimes forcing the rejection of one—is at the heart of the problems with transformative-experience decisions.

    “I think that that split is not resolvable in a straightforward way; and it's that split that's at the heart of some of these problems with decision-making and decision theory”

  • Calculating what one should do in purely third-personal consequentialist terms quickly produces 'robot morals'—sacrificing one life to save two on a simplistic calculus, never throwing a child's birthday party because the money should buy malaria bed nets—because it loses touch with the essential first-personal side of being human and thereby omits genuine values from the calculation.

    “If you calculate what you should do in kind of purely third-personal terms, you very quickly get into what I think of as like robot morals”

  • The paradox of choosing suffering—where people can rightly say a life filled with suffering is more valuable for them—is very difficult to explain on a quantifiable, person-independent calculus, but becomes intelligible once we admit the value of first-person experience, authenticity, and aesthetic value that cannot be reduced to an individual-independent number.

    “People can say that a life filled with suffering is actually more valuable for them. And, I think that to adequately explain that, let's say for someone who wants to try to reduce things to quantifiable characterizations, there's a little bit--it becomes very difficult.”

0.66

Evidence from averages and reference classes cannot be individualized to transformative decisions because experiential knowledge is epistemically inaccessible without the transformation itself.

5 pointscentrality 4/5
  • Because of how the brain works, some things—like what it is like to see red—cannot be learned through description or testimony and can only be known by having the experience directly; this epistemic transformation is illustrated by Mary's room, where a person raised in black and white learns something genuinely new upon first seeing color.

    “Sometimes experience teaches us things that we couldn't know in any other way.”

  • The reference class problem—knowing that evidence about others actually applies to people like you—takes on special significance in life-changing decisions because so many idiosyncratic variables beyond standard demographics matter, and because the high stakes mean getting the reference class wrong has severe consequences.

    “it seems like so many other variables matter other than the demographic variables that are usually assessed, that there is a kind of reference class problem here”

  • Even with the best available evidence, average values are misleading for high-stakes life decisions because your individual outcome could fall anywhere along a wide error bar—especially dangerous when that span extends from positive to negative—and you cannot fine-grain it to your own case.

    “because we're talking about high-stakes, life-altering decisions, the fact that you could fall anywhere on a span--like there can be a fairly large or significant error bar with respect to average values--can actually have a huge impact”

  • Normally we fine-grain average-value evidence by imaginatively projecting how we would respond to a situation, but in transformative experiences this is impossible because we lack the experiential information needed to perform that projection, so we cannot fix the reference class or individualize the data.

    “in this kind of context, we can't do it: because we lack the information that we need to be able to do that kind of imaginative projection”

  • When parents choose whether to give a deaf child cochlear implants versus raising them in the deaf community, making the choice by imaginatively comparing the child's life at 15 or 25 under each option is impossible, because each parent can experientially access only one of the two future lives, so the expected-value comparison the decision seems to require cannot actually be performed.

    “if you're supposed to make the decision by imagining what life would be like for your child as a member of the deaf community... and compare that to what it would be like for your child to have cochlear implants... Well, you can't possibly do that. You can't do both things.”