Transcript We Can Test the Experience Machine

Dan Weijers
Victoria University of Wellington
June 2011
 Lots of experimental philosophy is designed to test
empirical premises like: ‘We would not plug in’
 “Philosophers working in… ‘experimental philosophy’
have begun… to collect data about folk intuitions”

Nahmias, E., et al. (2007). Surveying Freedom: Folk Intuitions about Free Will and Moral
Responsibility, Philosophical Psychology, 18(5): 561.
 But lots also just wants to understand how judgements
about thought experiments are formed:
 “[To] use the methods of experimental psychology to
probe the way people make judgments that bear on
debates in philosophy”

Nadelhoffer, T. & Nahmias, E. (2007). The Past and Future of Experimental Philosophy,
Philosophical Explorations, 10(2): 123.
 Typical argument:
1.
2.
3.
If pleasure = g. good, then we’d plug in
We would not plug in
Therefore, pleasure ≠ g. good
 Test P2 (if true, go to next step, if false, then claim the
argument fails)
 ACP: you did a faulty test: the wrong ‘we’ or procedural bias
 Tweak supposedly irrelevant factor of thought experiment,
test again, and compare results (if different in right way,
claim that P1 is false because irrelevant factors unduly
influence judgements)
 ACP: You did a faulty test: the wrong ‘we’, or procedural bias,
or added other irrelevant factors with your tweak
 We can’t usefully test thought experiments with surveys
when…
 Respondents need to adopt the role of confronted
agents…
 i.e. when they have to predict what they would do if the
scenario were real
 Usually involves “confusion, incredulity, fear, etc.”

Smith, B. (2011). Can We Test the Experience Machine?, Ethical Perspectives, 18(1): 45.
 Because participants’ judgements about what they would
do would otherwise not be made from the correct frame of
mind… leading to the possibility that
 They might report what they think they should do or what the
experimenter wants to hear instead of what they would
actually do
 Our survey tests of thought experiments are less useful
to the extent that:
 Respondents need to adopt the role of confronted
agents
 When they “anticipate their own futures”, and/or
 They “identify with the moral decisions of others”
 Smith, B. (2011). Can We Test the Experience Machine?, Ethical Perspectives, 18(1): 46.
 The survey suffers from methodological weaknesses
 Sampling
 Participant ineptitude
 Procedural bias: prepping, wording, question order (Grice)
 “…the inverted experience machine, as well as
other[s]… have a unique set of characteristics that
make it impossible to gather the right subjects to test.
Therefore, in practice, these thought experiments are
impossible to test.”

Smith, B. (2011). Can We Test the Experience Machine?, Ethical Perspectives, 18(1): 37. (My
emphasis)
 “…unique set of characteristics…”:
 Requirement that participants adopt the role of confronted
agents
 When they “anticipate their own futures” – e.g. when asked ‘what
would you do?’
 Requirement for being a partially confronted agent: When they
“anticipate their own futures…, or attempt to identify with the
moral decisions of others, to some extent” (Smith, 2011, p 46)
 Setup makes it impossible for participants to adopt the role of
confronted agents
 Surveys “that address thought experiments about moral dilemmas”
(Smith, 2011, p 44) – Infer: all options involve significant losses?
 Participants opinion asked for via survey, rather than behaviour
observed in actual case (Smith, 2011, p 39 + elsewhere)
 Requirement that participants adopt the role of
confronted agents:
 When they “anticipate their own futures” (Smith, 2011, p
46) – e.g. ask ‘what would you do?’
 This requirement is not met
 ‘Anticipating their own futures’ is not a necessary aspect of
the Inverted Experience Machine
 E.g. What should a stranger choose in the Inverted Experience
Machine case?
 Requirement for being a partially confronted agent:
 When they “anticipate their own futures…, or attempt to
identify with the moral decisions of others, to some
extent” (Smith, 2011, p 46)
 This requirement is faulty
 Just because participants have to “attempt to identify with
the moral decisions of others” does not always mean that
they need to get (at all) emotional to give a useful response
 In fact making rational (not emotional) decisions is what we
expect of those making important decisions on behalf of
others in lots of cases
 E.g. policymakers – should uni be free?
 Amended requirement for being a partially
confronted agent:
 When they “anticipate their own futures…, or attempt to
identify with the moral decisions of others, to some
extent [in some cases]”
 This requirement does not apply to some cases of testing the
Inverted Experience Machine
 When deciding if someone should stick with a machine life
or go to reality we are better of putting emotions (“confusion,
incredulity, fear, etc.”) behind us.
 And, when trying to learn about what people think we should
really value in a life, the same goes
 It’s not clear where Smith stands on this
 If we can’t test the IEM because we don’t get
emotional, can philosophers make any use of it?
 Did anyone feel “confusion, incredulity, and fear etc.”
about the thought experiments?
 If not, then (according to Smith’s framework) your
judgement about it is useless
 Maybe Smith doesn’t mind this result – I’m not sure
 Smith 1: thought experiments that require confronted




agent respondents cannot be usefully surveyed
because the appropriate emotions don’t come up
Smith 2: Inverted Exp. Mach. requires confronted
agent respondents
Smith 3: Therefore, Inverted Exp. Mach. cannot be
usefully surveyed
Me: not all Inverted Exp. Mach.’s require confronted
agent respondents
Me: if Smith 1 is correct, then all use of dilemmas in
philosophy is a waste of time or at least less useful than
is usually thought