Sleeping Beauty in One or Many Worlds: Reclaiming the Halfer Position

The Sleeping Beauty Problem (SBP) has long been the "final boss" of subjective probability. For decades, the "Thirder" position—the belief that Beauty's credence in the coin landing Heads should be 1/3—has dominated the field. However, in this provocative new paper from the University of Oxford, Jiaxuan Zhang argues that the academic community has been misled. By bridging the gap between classical paradoxes and the Many-Worlds Interpretation (MWI) of quantum mechanics, Zhang provides a rigorous defense of the Halfer (1/2) position.

TL;DR

The paper argues that the transition from Sunday's objective 1/2 probability to Monday's subjective credence should not change because no "new" information is learned. It refutes the Thirder claims by exposing a series of logical shortcuts in their most famous arguments, proving that P(H)=1/2 is the only answer that satisfies both the Born rule in quantum mechanics and Bayesian consistency in classical probability.

The Core Conflict: Why 1/3 Seems Right (But Isn't)

The SBP is simple: Beauty is put to sleep. A fair coin is tossed.

• Heads: Woken once (Monday).
• Tails: Woken twice (Monday and Tuesday), with memory erasure in between.

Upon waking, what is her credence $P(H)$? Thirders argue that if this were repeated 1,000 times, she would wake up ~1,500 times total, with only 500 of those following "Heads." Thus, 1/3. Zhang identifies this as the Proportion Argument and strikes it down by distinguishing event weight from probability. Just because an event (Tails) creates more "instances" of waking doesn't mean the underlying probability of the coin toss has shifted.

Methodology: Fixing the Quantum and Classical Leak

1. The Quantum Correction

In the quantum version of SBP, the coin is a superposition $|\psi ⟩=\frac{1}{\sqrt{2}}(|H⟩+|T⟩)$. Some researchers previously assigned 1/3 by creating a four-event space and "deleting" the impossible one (Heads on Tuesday).

Zhang points out a renormalization error:

"In MWI, the modulus square of the amplitude... represents the probability of a certain world. There is only one split resulting from the single quantum coin toss."

By the Born rule, the "Heads World" and "Tails World" are equally weighted at 1/2. Any attempt to slice the "Tails World" into two days to reach 1/3 involves an unjustified second normalization.

2. Refuting Elga's Variant

Adam Elga's famous proof relies on the Principle of Indifference. Zhang uses a "Method (2')" variant to show that Elga's proof implicitly introduces information that isn't in the original problem.

As shown in the paper's derivation, when we properly apply Bayes' Theorem without Elga's extra assumptions, we find that the information available to Beauty actually supports $P(H|Mo)=2/3$ and $P(T|Mo)=1/3$, which perfectly reconciles with a total $P(H)=1/2$.

Critical Insight: The Technicolor Trap

The paper provides a brilliant takedown of Michael Titelbaum's Technicolor Beauty argument (where Beauty sees different colored papers on different days).

Titelbaum argued that seeing a specific color narrows the possibilities to 1/3. Zhang exposes the flaw: Overlapping Events.

• $P(Red)$ and $P(Blue)$ are NOT disjoint.
• In the Tails condition, she sees both.

When you account for the fact that these events overlap, the math collapses back to 1/2.

The Ultimate Test: Dutch Books

A "Dutch Book" is a set of bets that guarantees a loss if your credences are irrational.

• Causal Decision Theory (CDT): Treats "future you" as a separate agent.
• Evidential Decision Theory (EDT): Recognizes "future you" will likely act exactly as "current you" does.

Zhang argues that CDT is inappropriate for SBP because the "agent" on Tuesday is the same agent as Monday. When using EDT (the more logical choice for self-locating uncertainty), the Thirder position actually leads to a guaranteed loss, while the Halfer position remains safe.

Final Takeaway

This paper is a significant blow to the "Thirder" hegemony. It proves that:

1. MWI is robust: It doesn't need "quantum-only" probability rules to survive SBP.
2. Logic over intuition: The 1/3 answer relies on counting wakings, while 1/2 relies on counting truths.
3. Future Research: This framework can now be used to test other interpretations of quantum mechanics by seeing how they handle "subjective uncertainty in objective determinacy."

Conclusion: Next time you wake up and don't know what day it is, bet on 1/2. Your math (and your wallet) will thank you.