Other Minds
Last time we talked about what might be called
“global” skepticism– skepticism that anything at
all can be known.
We considered several responses to arguments
for skepticism: accepting skepticism, accepting
idealism, or simply denying that skeptical
arguments have any force (Moore’s response).
Nobody’s a Skeptic
Whatever we think about the particular
responses to skepticism, most contemporary
philosophers think that skepticism is neither
very compelling nor very worrisome.
We do know things, even if it’s difficult to say
how that’s possible. This is part of Moore’s
“common sense” legacy.
Local Skepticism
In addition to “global” skepticism there are
various “local” skepticisms: instead of claiming
that you can’t know anything at all, a local
skepticism says that you can’t know anything
about a certain topic.
For example, you might think that it was
impossible to know anything about God, even
though you could know lots of other things.
Global Skepticism
Global skepticism presents arguments that apply
generally to any claim. For example we saw that
Sextus Empiricus thought that there were equal
reasons for and against any claim. And Descartes
argued that if we were being deceived by an Evil
Demon, he could trick us about almost anything
(except that we existed and that we had the
experiences we did).
Local Skepticism
Local skepticism instead focuses on specific
features of the topic in question that are
supposed to make knowledge about that topic
For example, someone might claim that God is
unknowable, because She transcends reason
and logic, being their creator, or because the
human mind is too weak, or because we can’t
experience God directly.
Two Skepticisms
In today’s class and next Monday’s class we are
going to consider two different local skepticisms:
Skepticism about Other Minds: We can’t ever
know that other people have minds.
Skepticism about Induction: We can’t ever know
about unobserved phenomena.
The way we have of finding out about our own
minds is very different from the way we have of
finding out about other people’s minds.
To find out what I think, or feel, I might close my
eyes and concentrate on my thoughts or
sensations. This is called “introspection” which
just means “looking inward.”
However, I can’t “look inward” at your thoughts
and feelings. If I want to know what you think, I
have to observe how you act, and what you say.
The same is true if I want to know how you feel.
This would be silly in the first-person case. I
don’t have to look at how I act in order to find
out what I believe.
Who Has a Mind?
But does the fact that other people behave in
certain ways really justify us in believing that
they have minds?
You presumably think that some things don’t
have minds and some things do. Cats have
minds, and maybe fish do too, probably not
worms and definitely not bacteria, or rocks.
Who Has a Mind
But how do you know these things?
Panpsychists believe that everything has a mind,
including trees and rocks and gamma radiation
and empty regions of space.
“A rock doesn’t have a mind,” you might say, “it
doesn’t cry out when I kick it.” But the
panpsychist can just say: “I said it had a mind, I
didn’t say it had a mouth!”
Who Has a Mind?
Or how do you know that cats have minds?
Descartes, for instance, believed that only
humans had minds, because the mind = the
soul, and God had only given souls to humans.
“The cat cried out when you kicked it– isn’t that
proof that it feels pain?” “No,” Descartes would
say, “I could build a robot that did that, and the
robot wouldn’t feel pain.”
Philosophical Zombies
A philosophical zombie (as opposed to the
ordinary kind of zombie) is a thing that is in all
respects exactly like a person– it eats, and
sleeps, and drinks, and has conversations, and
goes on dates. But it has no “inner life”– there’s
nothing it’s like to be a zombie. It feels no pain
(even though it cries out when you strike it), and
it has no beliefs (even though it argues with you
forcefully about various subjects).
The Epistemological Problem
One way of stating the problem of other minds
is just this: how do you know that other people
aren’t philosophical zombies? What possible
argument can you give for attributing mental
states to them?
Local Skepticism
This isn’t global skepticism– the claim isn’t that
you can’t know anything. You might think that
knowledge = justified true belief. And you can be
justified in believing that you have hands, for
example, because you see them. If it’s true that
you have hands, then you know you have hands.
The problem of other minds asks: what justifies
you in believing other people have minds?
Unusual Justification
If there is a justification, it can’t be an ordinary
It’s not that you see, smell, taste, hear, or feel
other people’s minds.
And you don’t “directly” experience them either,
in the way you do your own.
The Conceptual Problem
In addition to the epistemological problem of
other minds (“How do we know?”) there is also
a conceptual problem of other minds: how do
we even think of other minds than our own? My
personal experiences allow me to think about
my own mind, and the appearances of physical
objects give me a way to think about them: but I
don’t experience other minds, and they don’t
cause appearances in me. How can I even think
the thought “You have a mind”?
The Conceptual Problem
That might sound a little weird. I know what it is
for me to have a mind. So when I think “she has
a mind” I am just saying that’s how it is for her.
But the skeptic thinks this is a verbal trick.
Wittgenstein famously remarked that it’s like
saying “I know what it is for it to be 5:00pm in
Hong Kong, therefore I know what it is for it to
be 5:00pm on the sun.”
Bertrand Russell (1872-1970)
Bertrand Russell was a British aristocrat (the 3rd
Earl Russell) who made lasting contributions to
logic, mathematics, and philosophy. Together
with G.E. Moore, he helped lead British
philosophical thought away from Absolute
Idealism towards a pluralistic realism.
Arguments from Analogy
Russell’s suggested solution to the problem of
other minds is an argument from analogy.
According to Russell, the argument takes this
Only A’s cause B’s
I observe a B_________________________
Therefore, I know that B was caused by an A.
Thunder and Lightning
Here’s an example that’s not about minds:
Only lightning causes thunder.
I hear some thunder in the distance (but I didn’t
see any lightning)._______________________
Therefore, there must have been some
unobserved lightning that was the cause of the
Analogy Argument for Other Minds
And here’s how the argument works for minds:
Premise 1: Only the sensation of thirst causes
water-drinking behavior.
Premise 2: That other person is drinking water
(but I don’t experience their thirst).________
Conclusion: Therefore, that person’s waterdrinking behavior was caused by an unobserved
New Analogy Argument
Russell doesn’t explicitly say this, but he seems
to suggest a different argument from analogy as
All A’s cause B’s.
I observe an A (but can’t observe B).
Therefore, there was an unobserved B
New Analogy Argument
Here’s how this might work for the mind:
Premise 1: Walking through the desert for four
hours with no water causes a sensation of thirst.
Premise 2: That person walked through the
desert for four hours with no water._________
Conclusion: That person is thirsty.
Both Arguments Together
And we can combine the two forms of argument
Premise 1: Walking in the desert causes thirst
AND thirst causes drinking water.
Premise 2: That guy walked in the desert AND
drank water.___________________________
Conclusion: Therefore, that guy was thirsty.
Evaluating the Argument
Premise 2 is obviously acceptable: as long as
we’re not being global skeptics, we’ll be inclined
to think that other people are in fact drinking
when it looks like they are, or that they walked
through the desert, when it looks like they did.
Russell’s Postulate
What can be said in defense of premise 1?
Russell proposes a postulate:
“If, whenever we can observe whether A and B
are present or absent, we find that every case of
B has an A as a causal antecendent, then it is
probable that most B’s have A’s as causal
antecedents, even in the cases where
observation does not enable us to know
whether A is present or not.”
Russell’s Postulate
So the idea is this. Only in my own case can I
ever observe both a sensation of thirst and
water-drinking behavior. (For others I can only
observe drinking, not thirst). But every time I
drank water, it was because I was thirsty.
Therefore, I can conclude that most water
drinking behavior is caused by sensations of
thirst, even though I can’t observe them in other
Induction on One Instance
Russell’s postulate seems to break down in
certain scenarios. Suppose you haven’t been
over to anyone else’s house, you’ve only been to
your own house. At your house, everyone takes
off their shoes at the door– nobody walks
around at home with their shoes on. It doesn’t
seem safe to conclude that at everybody else’s
house everyone takes off their shoes. You’d
need to look at more than just one house
(yours) to conclude that.
Induction on One Instance
Here’s another example. In the U.S., servers
come to your table to take your order, ask you if
you need refills, bring you the check, etc.– and
you never have to call them over. In fact,
Americans would probably think it was rude to
call across the restaurant to get service. Suppose
you’ve only ever lived in America. Can you
conclude that this is how it is in every other
country? It seems you’d need to look at more
than just one country to draw that conclusion.
Induction on One Instance
Russell wants us to conclude that certain
sensations are usually present in others when
they behave in certain ways, because that’s how
it usually is for us. But if I’ve only ever looked at
my mind, or my house, or my country, is that
enough to draw a conclusion about everyone’s
mind, or everyone’s house, or everyone’s
country? No, I don’t think it is.
Only Thirst
Russell needs to claim that thirst is the only
cause of drinking water. If there are other
causes, then you can’t infer from water-drinking
to sensations of thirst.
But aren’t there other causes? Sometimes you
need to cleanse your palate, or you’re trying to
prevent future thirst, or you’re just
absentmindedly drinking because it’s there.
Complexity Increases Certainty
Russell claims that the more complex the task is,
the better the analogical argument is. So if you
ask someone to solve a difficult mathematical
problem, and they give you the right answer,
you are safer in assuming that they did this in
the same way as you would.
But it would seem that the opposite is true. If I
give a difficult problem to 30 different computer
programmers, I’m liable to get 30 very different
programs written in 30 very different
programming languages. Why should we think
that there’s not an entirely different set of
complicated inner workings– not thoughts or
feelings, as for us– in other people? Maybe they
have a totally different complex way of
A lot of philosophers think that the argument
from analogy just doesn’t work, because only
looking at one instance (how things are for you),
is not a good way to generalize about how
things are in other instances, for other people.
We are justified in believing in other minds, but
that justification does not come from an
analogical argument.
The analogy viewpoint also leaves unsolved the
conceptual problem of other minds. Here’s what
Wittgenstein says:
“If one has to imagine someone else’s pain on
the model of one’s own, this is none too easy a
thing to do: for I have to imagine pain which I do
not feel on the model of the pain which I do
Ludwig Wittgenstein (1889-1951)
Ludwig Wittgenstein was the son of the second
wealthiest man in the Habsburg Empire, the
Austrian steel baron Karl Wittgenstein. His
father wanted his sons to be part of his steel
empire 3 (out of 5) of them committed suicide,
and Ludwig gave away his entire inheritance.
Wittgenstein was a student of Russell’s, and
many consider him the only philosophical genius
of the 20th Century.
Philosophical Confusion
Wittgenstein thought that the ordinary
analogical way we think of other people’s
mental states (as being ‘like’ ours) doesn’t make
He thought that once you cleared up the
confusion about what ‘That other person is in
pain’ really means, you would find that
obviously, we can know about other people’s
Next class when we study the problem of
induction, we’ll talk about a famous skeptic,
David Hume. But I’ll bring him up now just for
some background.
Hume thought there wasn’t a problem of other
minds, because he thought that there was no
evidence that I have a mind.
Hume against the ‘Self’
“For my part, when I enter most intimately into
what I call myself, I always stumble on some
particular impression or other, of heat or cold,
light or shade, love or hatred, pain or pleasure. I
never can catch myself at any time without a
perception, and never can observe anything but
the perception.”
Denial of Minds
What Hume denied was the picture on which
there were thinkers and thoughts, or perceivers
and perceptions: he argued that there were only
thoughts and perceptions, not minds that had
those things.
His argument was that when you introspected,
you found a bunch of thoughts, but you never
found you– the thing that does the thinking.
Wittgenstein only published one book and one
article in his lifetime, so most of what we know
about his views is from his notes and other
sources. G.E. Moore wrote about lectures
Wittgenstein gave in 1930-1933.
It Thinks
Moore writes, quoting Wittgenstein:
‘Just as no (physical) eye is involved in seeing, no
Ego is involved in thinking or in having
toothache’; and he quotes, with apparent
approval, Lichtenberg’s remark that ‘instead of
saying “I think” we ought to say “it thinks”’
The Cogito Again
The idea is that Descartes was wrong when he
concluded, “I think, therefore I am.” He thought
he was certain of a self– himself– a thinker that
did the thinking. But you don’t experience
yourself, you just experience your thoughts and
perceptions. You don’t see your eye when you
look out in the world. So the only thing
Descartes could really be sure of is not that he
exists, but that thoughts were happening.
Nothing to Analogize
What does this have to do with other minds? If
we deny that thinking is a relation between a
thinker and a thing thought, but instead
maintain that there is just “thinking happening”
or “no thinking happening,” then it is impossible
to say that I have certain thoughts and
experiences and so do other people. There are
no selves, so no other minds to do the thinking.
Ordinary Language Philosophy
Wittgenstein thought that philosophical
confusions like this could be resolved by
investigating how we use language and what
purpose it is put to.
He asks why we use mental language like ‘he is
in pain’ or ‘he thinks that it will rain later’. If we
can figure that out, we’ll know what these
assertions really mean.
The Role of Mental Talk
In Philosophical Investigations, he says:
“A child has hurt himself and he cries: and the
adults talk to him and teach him new
exclamations and, later, sentences. They teach
the child new pain-behavior… the verbal
expression of pain replaces crying and does not
describe it [the pain].”
Here, someone saying something like “I am in
pain” shows that they are in pain in much the
same way that crying does– but it does not
describe the state that they are in (crying
doesn’t describe your pain, it shows it). And
saying ‘he is in pain’ is a lot like saying ‘he is
crying’– you are saying that he is engaging in
certain pain-behavior.
Criterial Connections
Wittgenstein’s idea was that certain behavior
was criterial for the application of mental talk–
that it provided a standard for when to use that
talk appropriately. When you learn the language
game involving ‘pain’ you learn what behaviors
make it appropriate to say that someone else is
in pain. And thus saying that they are in pain
(correctly) is simply applying ‘pain’ talk when
you notice the appropriate behaviors.
Behaviorist Solution
Therefore, there is no problem of other minds.
So long as we are not global skeptics, and we
accept that we can know when other people are
crying or clutching their sides, or groaning, or
saying “I am in pain” then we can know that
they are in pain… what it is for them to be in
pain is that they exhibit such behaviors.
Behaviorism is the idea that mental states are
not really “inner” things at all, but rather
dispositions to behave in certain ways. To say
“Michael believes that Jamaica is a nice vacation
spot” is to say something like, “Michael is
inclined to buy a ticket to Jamaica when he has
free time and disposable income.”
The behaviorist has an obvious solution to the
problem of other minds: if mental states =
behavior, then we can observe people’s mental
states by observing their behavior.
Old Joke: Two behaviorists meet on the street.
“You’re feeling good today,” says one, “how am I
Objection to Behaviorism
The most serious objection to behaviorism is
that it can’t be true, because it requires a 1-to-1
connection between behaviors and mental
states, when in reality any given mental state
can bring about a range of different behaviors,
depending on the other mental states involved.
Objection to Behaviorism
For example, I believe that Jamaica is an
excellent vacation spot. Does this mean that I
will buy a plane ticket there over winter break?
Well, only if I believe that a plane is the best way
to get there, and believe that the airline will
honor my ticket, and desire to take a vacation,
and don’t have a stronger desire to do
something else (visit family), etc.
Behavior Indirectly Connected with
Mental States
Your beliefs and desires are only indirectly
related to your behavior. Any particular behavior
is consistent with your having a large number of
different beliefs and desires.
Why did you take your umbrella? Because you
thought it was raining and you didn’t want to
get wet. Or because you thought it was sunny
and you didn’t want to get burned.
Or because…
Or because you thought there were bears
outside and they are afraid of umbrellas.
Or because you wanted a walking stick, and the
umbrella was the best you had.
Or because you thought the umbrella was a
sandwich and you were hungry…
Behavior ≠ Mental States
You can never make an identification between
behavior B and mental state M: B = M.
The reason is that people’s behavior depends on
all their beliefs and desires, not just one, M.
By changing mental states other than M, and
keeping M fixed, you can change the behavior.
A denial of the existence of thinkers and
perceivers led Wittgenstein to claim that
understanding mental states by analogy with
our own case made no sense. He proposed an
alternative, behaviorist theory on which mental
states were identified with certain behaviors.
Although behaviorism of this form was popular
for a while, most philosophers reject it, because
there is no nice correspondence between
mental states and behaviors.
The Problem
Here’s how I introduced the problem of other
minds: we can find out about the physical world
and the states of other people’s bodies
(behaviors) by perceiving them– looking and
touching. We can find out about our own mental
states via introspection– “looking inward.” But
how can we find out about other minds?
But there are lots of things that we are justified
in believing in which we can neither see nor
For example, I believe in electrons and black
holes, even though I can’t see either of them.
And the science behind these things really does
justify my beliefs– I know that there are
electrons and black holes.
One popular solution to the problem of other
minds is to view it as an abductive argument: an
inference to the best explanation. We observe a
complicated range of effects– people’s
behavior– and we derive a model which, if it
were true, it would explain people’s behavior.
Since there is no better model that explains the
behavior, we accept this model as true: likely,
people do have pains, beliefs, and desires.
Example: Neptune
Before the discovery of Neptune, in the 19th
Century, measurements of the orbit of Uranus
showed that it did not behave in the way that it
should, if there are no other planets in the solar
system. Astronomers then inferred that the best
explanation for its irregular orbit was that there
was another undiscovered, unobserved planet.
Later, the planet was observed, and we call it
Reverse Engineering
In general, abduction reasons from the effects to
their most likely causes. It is like “reverse
engineering.” For example, if you want to find
out how a particular piece of software works,
you give it various inputs and see how it
behaves– does it take a long time to return
answers to certain problems or does it crash?
Observing its behavior allows you to infer the
hidden structure of its program.
Reverse Engineering the Mind
In the case of other people’s minds, we observe
inputs (what is in people’s line of sight, what’s in
range of hearing, what’s in their mouth…) and
we observe outputs– how they behave, what
they say. We then infer the hidden, internal
processes that mediate between input and
output: beliefs, desires, emotions, and so on.
Computational Theory of Mind
In fact, if the computational theory of mind is
correct, then inferring the mental states of other
people is the exact same thing as reverse
engineering a piece of software. According to
CTM, the mind is the software of the brain. So
observing the inputs and outputs of that
software allows us to infer the nature of the
software itself: the beliefs and desires of other
Comparison with Analogy
This strategy is different from the argument
from analogy. There, it was important to find out
how our own minds worked, and then to reason
that other people must be like us.
The abductive inference strategy looks at
observable inputs and outputs for lots of other
people. We aren’t just restricted to our own
Comparison with Behaviorism
It is different from Behaviorism too. The
behaviorist wanted to identify mental states
with people’s outputs: their behavior. On the
abductive inference strategy, behavior is
different fromthe hidden, internal mental states,
but it is evidence regarding what those states
are like.
But how exactly are beliefs and desires
supposed to explain the behavior of other
Consider this case. I call a friend in the U.S. and
ask him to give a guest lecture two weeks from
now in class. She says yes, she’d love to. Where
will she be two weeks from now?
All our sophisticated knowledge of brain science
won’t allow us to make a good prediction here.
We could scan her brain with PET scans and
fMRI’s for days, record her individual neurons
with invasive electrodes, and measure every
aspect of her physiology… but still, that wouldn’t
come close to a reasonable prediction of where
she’ll be in two weeks. But we know the
answer… she’ll be here!
The Abductive Argument
1. My friend says “I will come teach class at HKU
two weeks from now.
2. The best explanation for this is that she
intends to fly to HKU and teach class (she’s
not lying).
3. People generally do what they intend to do.
4. Therefore she will fly to HKU and be in class
in two weeks.
Confirmed by Experience
The fact that such prediction is so often
successful, and so much superior to all of our
other sophisticated science, is good reason to
believe that the theoretical entity it posits– my
friends intention– and the causal relations it
assumes that entity to enter into– intentions to
do X cause people to do X– are real.
Mind Reading
There is evidence that the way we actually go
about attributing mental states to other people
and explaining and predicting their behavior
involves these sorts of abductive inferences.
Three-year old children and people who are high
on the autism spectrum have trouble attributing
beliefs to other people, and so have trouble
making accurate predictions about others.
The Sally-Ann Task
Young children know where the cookie is (in the
box), so they assume other people know where
it is. They don’t have the idea that other people
can have beliefs that differ from their own.
Mind Reading
As we mature, we become more adept at
positing distinct arrays of beliefs and desires in
other people, and use those models in a
predictive and explanatory way.
This is proof that evolution itself has seen fit to
adopt the “theoretical entities” approach to
mental states, and treat them as unobserved
causes of others’ behaviors.
There are three objections to the abductive
First, the theory is circular: it needs to assume
its own truth to prove that it’s true.
Second, we have addressed the epistemological,
but not the conceptual, problem of other minds.
Third, we still have no reason to suppose that
other people have conscious inner lives.
Hyslop (SEP) argues that the functionalist
solution to the problem of other minds is
circular: functionalism is a theory about how
everyone’s mind works. To establish that theory,
you have to have lots of knowledge about
minds– yours and other people’s– so you have
to solve the problem of other minds before you
can use functionalism to solve the problem of
other minds.
Missing the Point
I think this misses the point of the abductive
strategy. Here, we do assume in advance that
functionalism is true, and see what that
predicts. But then we go out looking and we find
that our predictions have come true. So we
conclude that functionalism must be true,
because no other theory better explains the
The Conceptual Problem
The conceptual problem of other minds is that
we cannot even have an idea of someone else
having a mind, because we can neither see their
mental states, nor introspect them, and those
are the only ways we have of conceptually
grasping other things.
Easy Response
In this simple form, I think the abductive
strategy/ functionalism has an easy response.
We can have ideas of things we don’t see or
introspect, like electrons, black holes, and even
Neptune (at one point in history). The model of
concepts proposed by the “conceptual problem”
is deficient.
Wittgenstein’s Conceptual Problem
However, I think there is a more difficult
problem if the considerations urged by
Wittgenstein are correct. If there really are no
thinkers, but only thoughts, how do I think of a
thought that is not thought by me, but by
someone else? Thoughts aren’t thought by
Response to Wittgenstein
I think we can respond to the challenge though:
Wittgenstein says there is no eye in the field of
vision. We see things, we don’t see an eye
seeing things. Similarly, there is no thinker in
what is thought. When I introspect, I see
thoughts, but there is no thinker there among
the thoughts.
Response to Wittgenstein
But this is no reason to deny that eyes exist. You
see the outside world, not an eye seeing the
outside world. But you do it with your eyes.
They still exist, even if they’re not what you see.
Similarly, thinkers exist. The reason Hume
doesn’t see himself among his thoughts is that
his self is not a thought, it’s the thing having
them. This is no reason to think it doesn’t exist.
Philosophical Zombies Again
The big problem, as I see it, with the abductive
strategy/ functionalism is that it doesn’t give us
any reason to suppose that other people have
conscious inner lives.
If CTM is true, other people are computers that
run sophisticated programs. But computers
aren’t all conscious, or so we think. What reason
do we have to think those computers are
The Problem of Other Minds
The epistemological problem of other minds is a
variety of local skepticism. It’s the worry that we
can’t have knowledge about other minds
because we don’t ever perceive or experience
the thoughts of others.
Three Solutions
We considered three potential solutions:
1. Analogy: we can know about other minds by
analogy with our own mind.
2. Behaviorism: we can know about other
minds because mental states = behavior, and
we can observe behavior.
3. Abduction: we can know about other minds
the way we know about other unobserved
entities: inference to the best explanation.
The main objection to the argument from
analogy was that it relied on too few instances
(only one). Just because I know how something
is in my mind/ at my house/ in my country,
doesn’t mean I know how it is in every mind/ at
every house/ at every country.
The main objection to behaviorism was that,
while it provided a nice solution to the problem,
it was implausible to believe that mental states
were just dispositions to behave in certain ways.
One and the same belief could cause any
behavior you can imagine, given enough other,
different mental states.
It’s quite plausible that the way we know about
other people’s beliefs and desires is by inferring
them as the most reasonable explanation for
their behavior. However, even if we accept this
story, we still have no account of how we know
that other people have conscious inner lives.

Other Minds - Michael Johnson's Homepage