infinite observations and irrationality
There are two interesting features of theoretical models of mindreading
which have consequences for understanding mindreading in practice.
Think of decision theory: calculating any subjective preference or utility requires
an infinite number of observations.
If, like Davidson, you think that decision theory provides a model for of mental state attribution,
then the mindreading in theory also requires an infinite number of observations.
Of course I am not suggesting that this is a practical problem.
As I noted at the start, people frequently agree on attributions and are confident in them.
In all the attempts to characterise how mindreading could work in principle that I know,
axioms of rationality are required to link observed behaviour to mental states.
In practice, of course, people are only rational within narrow bounds.
Research in decision theory is a good source of situations where people
consistently make choices that appear irrational.
The Ellsberg Paradox, for example.
(https://philosophical-issues-in-behavioural-science.butterfill.com/docs/lecture_03/ellsberg_paradox/)
Because there are many ways you could compensate for irrationality,
there are many ascriptions of mental states which will all fit the evidence equally well.
∴ background assumptions needed
= ‘mindreading standards’
In any practical case, mindreading will depend on a set of background
assumptions, which I will pre-emptively call ‘mindreading standards’.
(The label ‘standards’ will be justified later.)
objections
- perceiving mental states
- foundational platitudes (Heider, 1958)
Foundational heuristics: if Ayesha tells me that she is angry,
she is angry.
∞todo: say something about Roth & Marsella (2026) (Glasgow source)
I’m happy to talk about these lines of objection in discussion,
but I’m aware that there are a range of views here and I do not
have novel persuasive arguments to offer today.
What I will say is just that, as I see things, neither of these
objections will work unless you can
extending the argument for background conditions
+ processing costs (Roth & Marsella, 2026)
Processing costs: mundane mindreading needs to be fluent and often occurs
in the context of conversation or joint activities, so even if decisive evidence
were available, acquiring and processing it would often be impractical.
[∞todo link to Roth & Marsella (2026) as Glasgow source]
