In causal decision theory, the perspective/​indexical aspect could lead to reflective inconsistency. This is generally regarded as a problem. Here I present a thought experiment to show why this view may require further review.
Two Anaesthetics
Suppose you are about to undergo a major operation. You can choose one of the following two anesthetics:
Drug A functions as follows:
It paralyzes your body and relaxes your muscles
It prevents long-term memory from forming for the duration of the operation
Drug B functions as follows:
It paralyzes your body and relaxes your muscles (same as A)
It renders you unconscious for the duration of the operation
Suppose the two drugs are equal in all other considerations such as safety, long-term side effects, etc. But Drug A is covered by your health insurance while Drug B requires you to pay 1 dollar out-of-pocket. What would you pick?
Reflective Inconsistency
Drug A would lead to the dreaded anesthesia awareness, that you will feel the excruciating pain and trauma of being operated on. To prevent it from happening, for the mere cost of 1 dollar, Drug B is the obvious choice.
Yet for moments after the operation, you would be wishing that you had chosen Drug A instead. The choice of drugs does not lead to any long-term difference, other than the fact that you would be 1 dollar richer if you had gone with Drug A. Furthermore, you know this before the operation.
If the later you dictate your current decision it would mean Drug A is the clear winner. (Consider this as a case of post-commitment, in contrast to pre-commitments where an earlier self takes away the decision at a later point, often appearing in decision-making problems such as Parfit’s Hitchhiker) Hence the inconsistency.
Decision Theories
CDT has a naturally built-in indexical element. Considering the fact that the decision is based on a pre-operation perspective, it will choose Drug B over A.
However, for non-indexical decision theories, such as UDT or EDT, the approach would be less clear, and it could very well lead to choosing Drug A. Should it only consider the utility at the end of the experiment? Or should it get around the indexicals with some self-sampling assumptions like common camps in anthropics? Neither approach seems problem-free.
Which Anaesthetic To Choose?
In causal decision theory, the perspective/​indexical aspect could lead to reflective inconsistency. This is generally regarded as a problem. Here I present a thought experiment to show why this view may require further review.
Two Anaesthetics
Suppose you are about to undergo a major operation. You can choose one of the following two anesthetics:
Drug A functions as follows:
It paralyzes your body and relaxes your muscles
It prevents long-term memory from forming for the duration of the operation
Drug B functions as follows:
It paralyzes your body and relaxes your muscles (same as A)
It renders you unconscious for the duration of the operation
Suppose the two drugs are equal in all other considerations such as safety, long-term side effects, etc. But Drug A is covered by your health insurance while Drug B requires you to pay 1 dollar out-of-pocket. What would you pick?
Reflective Inconsistency
Drug A would lead to the dreaded anesthesia awareness, that you will feel the excruciating pain and trauma of being operated on. To prevent it from happening, for the mere cost of 1 dollar, Drug B is the obvious choice.
Yet for moments after the operation, you would be wishing that you had chosen Drug A instead. The choice of drugs does not lead to any long-term difference, other than the fact that you would be 1 dollar richer if you had gone with Drug A. Furthermore, you know this before the operation.
If the later you dictate your current decision it would mean Drug A is the clear winner. (Consider this as a case of post-commitment, in contrast to pre-commitments where an earlier self takes away the decision at a later point, often appearing in decision-making problems such as Parfit’s Hitchhiker) Hence the inconsistency.
Decision Theories
CDT has a naturally built-in indexical element. Considering the fact that the decision is based on a pre-operation perspective, it will choose Drug B over A.
However, for non-indexical decision theories, such as UDT or EDT, the approach would be less clear, and it could very well lead to choosing Drug A. Should it only consider the utility at the end of the experiment? Or should it get around the indexicals with some self-sampling assumptions like common camps in anthropics? Neither approach seems problem-free.